diff --git a/benchmarks/benchmarks/scalability/results/expA_gpu.jsonl b/benchmarks/benchmarks/scalability/results/expA_gpu.jsonl new file mode 100644 index 0000000..23f339f --- /dev/null +++ b/benchmarks/benchmarks/scalability/results/expA_gpu.jsonl @@ -0,0 +1,134 @@ +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (1x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.751781+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (1x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.767008+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (1x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.782660+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (1x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.801008+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (1x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.815779+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (2x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.838134+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (2x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.858610+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (2x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.875171+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (2x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.894646+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (2x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.915044+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (4x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.942859+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (4x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:05.970720+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (4x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.007558+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (4x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.034722+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (4x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.062636+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (8x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.107562+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (8x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.153434+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (8x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.207524+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (8x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.256314+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (8x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.302434+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (16x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.378256+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (16x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.470563+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (16x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.553291+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (16x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.645453+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (16x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.760901+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.782845+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.801480+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.821504+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.839897+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.857280+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.878748+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.900802+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.921641+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.943913+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.965986+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:06.998425+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.028388+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.058070+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.086977+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.117412+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.166195+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.218128+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.262568+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.307473+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.351452+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.434913+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.520984+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.610095+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.700839+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:07.788975+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.396778+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.441687+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.485612+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.534802+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.579670+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.636398+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.690741+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.745550+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.799934+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.854460+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.925416+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:08.994186+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:09.065116+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:09.136865+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:09.218059+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:09.317589+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:09.415516+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:09.521126+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:09.621731+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:09.723788+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:09.890853+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:10.055728+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:10.222333+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:10.395936+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "07f1b6f9ac4dbbbec2ef88058264b0f12b7b9487"}, "timestamp_utc": "2026-11-04T00:29:10.562965+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (1x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.761578+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (1x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.775522+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (1x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.792823+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (1x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.808020+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (1x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.824275+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (2x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.845170+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (2x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.865368+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (2x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.884332+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (2x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.902060+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (2x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.919362+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (4x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.942430+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (4x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:10.970461+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (4x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.005753+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (4x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.034282+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (4x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.061960+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (8x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.102563+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (8x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.142858+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (8x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.189325+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (8x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.230407+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (8x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.277908+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (16x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.358603+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (16x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.452639+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (16x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.536629+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (16x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.625719+00:00"} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 670, in run_collection\n return run_zerorl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 354, in run_zerorl_collection\n trainer.rollout_phase()\n ~~~~~~~~~~~~~~~~~~~~~^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/train.py\", line 163, in rollout_phase\n outputs = env_step(self.env, self.agent, state_processed)\n File \"/home/research/projects/zeroRL/zerorl/functions.py\", line 55, in env_step\n outputs: dict[str, Tensor] = agent.get_action(state) #type: ignore[operator]\n ~~~~~~~~~~~~~~~~^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/utils/_contextlib.py\", line 124, in decorate_context\n return func(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 95, in get_action\n logits, value = self.forward(state, **kwargs)\n ~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/zerorl/helpers/factory.py\", line 162, in forward\n x = self.extract_layer(state)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 shapes cannot be multiplied (16x348 and 8x64)\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.729209+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.748711+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.766339+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.783685+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.801120+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.818081+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.839884+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.861573+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.883628+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.905772+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.926227+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.956934+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:11.987849+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.018536+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.049055+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.077354+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.126444+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.175415+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.226220+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.277507+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.329612+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.416790+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.502061+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.590825+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.677847+00:00"} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "Traceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 672, in run_collection\n return run_sb3_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 437, in run_sb3_collection\n load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed))\n ~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/benchmarks/scalability/policy_reference.py\", line 240, in load_into_sb3\n getattr(layer, attr).copy_(sd[ref_key])\n ~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^\nRuntimeError: The size of tensor a (348) must match the size of tensor b (8) at non-singleton dimension 1\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:12.762190+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:13.436525+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:13.483025+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:13.529223+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:13.579426+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:13.625200+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:13.680795+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:13.734415+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:13.788991+00:00"} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "failed", "phases": {"init_seconds": 0.0, "warmup_seconds": 0.0, "measure_seconds": 0.0}, "transitions": {"warmup_target": 16000, "warmup_actual": 0, "measure_target": 100000, "measured": 0, "exact": false}, "metrics": {}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000}, "error": "RuntimeError: TensorDictModule failed with operation\n _TorchRLCollectPolicy(\n (trunk): TorchRLSharedTrunk(\n (trunk): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (actor_head): Linear(in_features=64, out_features=4, bias=True)\n (critic_head): Linear(in_features=64, out_features=1, bias=True)\n (actor_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=4, bias=True)\n )\n (critic_path): Sequential(\n (0): Sequential(\n (0): Linear(in_features=8, out_features=64, bias=True)\n (1): Tanh()\n (2): Linear(in_features=64, out_features=64, bias=True)\n (3): Tanh()\n )\n (1): Linear(in_features=64, out_features=1, bias=True)\n )\n )\n )\n in_keys=['observation']\n out_keys=['action', 'log_prob', 'entropy', 'value'].\n\nThe above exception was the direct cause of the following exception:\n\nTraceback (most recent call last):\n File \"/home/research/projects/zeroRL/benchmarks/scalability/run.py\", line 148, in main\n art = run_collection(framework, cfg, num_envs, seed,\n device=args.device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 674, in run_collection\n return run_torchrl_collection(cfg, num_envs, seed, device=device)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 559, in run_torchrl_collection\n collector1 = Collector(\n wrapper, tdm, backend=\"direct\", trust_policy=True,\n ...<2 lines>...\n **_collector_device_kwargs(device),\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 253, in __call__\n return super().__call__(*args, **kwargs)\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 869, in __init__\n self._maybe_make_final_rollout(make_rollout=self._use_buffers)\n ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torchrl/collectors/_single.py\", line 1566, in _maybe_make_final_rollout\n policy_output = self._wrapped_policy(policy_input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 328, in wrapper\n return func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/utils.py\", line 372, in wrapper\n result = func(_self, tensordict, *args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1238, in forward\n raise err from RuntimeError(\n f\"TensorDictModule failed with operation\\n{module}\\n{in_keys}\\n{out_keys}.\"\n )\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1210, in forward\n raise err\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1194, in forward\n tensors_out = self._call_module(tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/tensordict/nn/common.py\", line 1153, in _call_module\n out = self.module(*tensors, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/benchmarks/scalability/collection.py\", line 652, in forward\n features = self.trunk.trunk(obs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/container.py\", line 253, in forward\n input = module(input)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1778, in _wrapped_call_impl\n return self._call_impl(*args, **kwargs)\n ~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/module.py\", line 1789, in _call_impl\n return forward_call(*args, **kwargs)\n File \"/home/research/projects/zeroRL/.venv/lib/python3.13/site-packages/torch/nn/modules/linear.py\", line 134, in forward\n return F.linear(input, self.weight, self.bias)\n ~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^\nRuntimeError: mat1 and mat2 must have the same dtype, but got Double and Float\n", "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "dc8015dc59d58c7ce8bc9b57f5d6ea975a99875a"}, "timestamp_utc": "2026-11-04T12:20:13.846170+00:00"} diff --git a/benchmarks/exp_A_gpu.jsonl b/benchmarks/exp_A_gpu.jsonl new file mode 100644 index 0000000..283d4f3 --- /dev/null +++ b/benchmarks/exp_A_gpu.jsonl @@ -0,0 +1,75 @@ +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.21089445799589157, "warmup_seconds": 26.083964532939717, "measure_seconds": 157.45702664996497, "boundary_seconds": 0.0003071138635277748}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 635.0939181793717, "init_seconds": 0.21089445799589157, "warmup_seconds": 26.083964532939717, "measure_seconds": 157.45702664996497}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 64, "rollout_calls_measure": 400, "buffer_capacity_rows_per_env": 116000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T22:37:56.695799+00:00", "extra": {"fingerprint": {"first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031], "actions_head": [0.000782, -0.001812, -0.002553, -6.2e-05, -0.005506, 0.007999, 0.009535, 0.003281, 0.00383, 0.007872, 0.005056, 0.008516, 0.010834, 0.002244, -0.001553, 0.000921, 0.005364, -0.003047, 0.000427, 0.006886], "action_sum_warmup": 43.95182722521713, "action_sum_measure": 5341.71044840978}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.014084093039855361, "warmup_seconds": 24.91296833800152, "measure_seconds": 156.03731594700366, "boundary_seconds": 0.00027909502387046814}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 640.8723412927961, "init_seconds": 0.014084093039855361, "warmup_seconds": 24.91296833800152, "measure_seconds": 156.03731594700366}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 64, "rollout_calls_measure": 400, "buffer_capacity_rows_per_env": 116000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T22:40:57.770369+00:00", "extra": {"fingerprint": {"first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669], "actions_head": [0.000776, -0.001994, -0.002623, 7.4e-05, -0.005503, 0.007947, 0.009494, 0.003432, 0.003652, 0.00791, 0.005274, 0.008565, 0.010683, 0.002282, -0.001344, 0.00097, 0.005429, -0.001209, -1.5e-05, 0.001186], "action_sum_warmup": 43.855298583221156, "action_sum_measure": 5336.224498329553}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.03953766613267362, "warmup_seconds": 25.21254638209939, "measure_seconds": 154.49796889908612, "boundary_seconds": 0.00029807700775563717}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 647.2576999722066, "init_seconds": 0.03953766613267362, "warmup_seconds": 25.21254638209939, "measure_seconds": 154.49796889908612}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 64, "rollout_calls_measure": 400, "buffer_capacity_rows_per_env": 116000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T22:43:57.630739+00:00", "extra": {"fingerprint": {"first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488], "actions_head": [0.000694, -0.002019, -0.00267, -1e-05, -0.005637, 0.007996, 0.009454, 0.003408, 0.003777, 0.007971, 0.005104, 0.008526, 0.01072, 0.002231, -0.001511, 0.000946, 0.005334, -0.000844, 0.002764, 0.002006], "action_sum_warmup": 43.93581202189671, "action_sum_measure": 5342.967553191695}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.01776214106939733, "warmup_seconds": 24.280628454871476, "measure_seconds": 156.35257879691198, "boundary_seconds": 0.0002863919362425804}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 639.5801129055317, "init_seconds": 0.01776214106939733, "warmup_seconds": 24.280628454871476, "measure_seconds": 156.35257879691198}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 64, "rollout_calls_measure": 400, "buffer_capacity_rows_per_env": 116000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T22:46:58.390453+00:00", "extra": {"fingerprint": {"first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156], "actions_head": [0.000717, -0.001973, -0.002607, 4e-06, -0.005605, 0.008015, 0.009436, 0.00335, 0.003709, 0.007889, 0.005196, 0.008655, 0.010753, 0.002205, -0.001481, 0.000933, 0.005331, 0.001077, 0.004181, 0.000296], "action_sum_warmup": 43.957922175046406, "action_sum_measure": 5337.2855851439745}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.019599749008193612, "warmup_seconds": 25.28770361491479, "measure_seconds": 155.58277338603511, "boundary_seconds": 0.0003206320106983185}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 642.7446806843967, "init_seconds": 0.019599749008193612, "warmup_seconds": 25.28770361491479, "measure_seconds": 155.58277338603511}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 64, "rollout_calls_measure": 400, "buffer_capacity_rows_per_env": 116000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T22:49:59.391056+00:00", "extra": {"fingerprint": {"first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815], "actions_head": [0.000724, -0.001868, -0.00252, 0.000102, -0.005545, 0.007989, 0.00947, 0.003329, 0.003748, 0.007969, 0.005124, 0.008495, 0.010744, 0.002225, -0.00142, 0.000975, 0.005403, 0.003445, 0.005474, -0.000733], "action_sum_warmup": 43.35931286640698, "action_sum_measure": 5333.122217774755}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.025503036100417376, "warmup_seconds": 17.47718825004995, "measure_seconds": 108.99492171988823, "boundary_seconds": 0.0004224099684506655}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 917.4739375197246, "init_seconds": 0.025503036100417376, "warmup_seconds": 17.47718825004995, "measure_seconds": 108.99492171988823}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 32, "rollout_calls_measure": 200, "buffer_capacity_rows_per_env": 58000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T22:52:05.997155+00:00", "extra": {"fingerprint": {"first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031], "actions_head": [0.000785, -0.001813, -0.002554, -6.1e-05, -0.005508, 0.007999, 0.00953, 0.003282, 0.003827, 0.007871, 0.005057, 0.008517, 0.010833, 0.002242, -0.001557, 0.00092, 0.005365, 0.000772, -0.001816, -0.002576], "action_sum_warmup": 43.27094245591434, "action_sum_measure": 5339.617813282213}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.021516581997275352, "warmup_seconds": 17.509435883024707, "measure_seconds": 111.8401317379903, "boundary_seconds": 0.000577263068407774}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 894.1334246124784, "init_seconds": 0.021516581997275352, "warmup_seconds": 17.509435883024707, "measure_seconds": 111.8401317379903}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 32, "rollout_calls_measure": 200, "buffer_capacity_rows_per_env": 58000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T22:54:15.504211+00:00", "extra": {"fingerprint": {"first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669], "actions_head": [0.000779, -0.001996, -0.002624, 7.6e-05, -0.005505, 0.007947, 0.009489, 0.003434, 0.00365, 0.00791, 0.005276, 0.008565, 0.010682, 0.002281, -0.001348, 0.00097, 0.005429, 0.000745, -0.001912, -0.002532], "action_sum_warmup": 43.22573202713829, "action_sum_measure": 5344.897832285389}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.020272323861718178, "warmup_seconds": 17.240991229191422, "measure_seconds": 110.6645437979605, "boundary_seconds": 0.0004209189210087061}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 903.6317917920423, "init_seconds": 0.020272323861718178, "warmup_seconds": 17.240991229191422, "measure_seconds": 110.6645437979605}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 32, "rollout_calls_measure": 200, "buffer_capacity_rows_per_env": 58000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T22:56:23.538369+00:00", "extra": {"fingerprint": {"first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488], "actions_head": [0.000698, -0.002021, -0.00267, -8e-06, -0.005638, 0.007997, 0.009449, 0.003408, 0.003774, 0.00797, 0.005105, 0.008526, 0.01072, 0.002229, -0.001515, 0.000946, 0.005336, 0.000815, -0.001899, -0.002512], "action_sum_warmup": 43.5179678819768, "action_sum_measure": 5333.288119125486}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.02052238886244595, "warmup_seconds": 17.51575354207307, "measure_seconds": 110.35184138081968, "boundary_seconds": 0.0004116971977055073}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 906.1924001331713, "init_seconds": 0.02052238886244595, "warmup_seconds": 17.51575354207307, "measure_seconds": 110.35184138081968}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 32, "rollout_calls_measure": 200, "buffer_capacity_rows_per_env": 58000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T22:58:31.534589+00:00", "extra": {"fingerprint": {"first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156], "actions_head": [0.000719, -0.001974, -0.002608, 5e-06, -0.005607, 0.008015, 0.009431, 0.00335, 0.003707, 0.007888, 0.005197, 0.008656, 0.010754, 0.002205, -0.001483, 0.000932, 0.005332, 0.000732, -0.001744, -0.002633], "action_sum_warmup": 43.26587468172511, "action_sum_measure": 5336.730076935663}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.022692446131259203, "warmup_seconds": 17.897073023952544, "measure_seconds": 111.59524024208076, "boundary_seconds": 0.0004096140619367361}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 896.0955662900362, "init_seconds": 0.022692446131259203, "warmup_seconds": 17.897073023952544, "measure_seconds": 111.59524024208076}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 32, "rollout_calls_measure": 200, "buffer_capacity_rows_per_env": 58000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:00:41.156541+00:00", "extra": {"fingerprint": {"first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815], "actions_head": [0.000727, -0.00187, -0.002521, 0.000104, -0.005546, 0.007989, 0.009465, 0.00333, 0.003745, 0.007968, 0.005126, 0.008495, 0.010742, 0.002223, -0.001424, 0.000974, 0.005404, 0.000728, -0.0019, -0.002511], "action_sum_warmup": 43.05671322600756, "action_sum_measure": 5332.0931705472}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.06100778607651591, "warmup_seconds": 13.44532627100125, "measure_seconds": 85.18653371697292, "boundary_seconds": 0.0005897171795368195}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1173.894460035713, "init_seconds": 0.06100778607651591, "warmup_seconds": 13.44532627100125, "measure_seconds": 85.18653371697292}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 16, "rollout_calls_measure": 100, "buffer_capacity_rows_per_env": 29000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933, 2822786858, 3633331186]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:02:19.969662+00:00", "extra": {"fingerprint": {"first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031], "actions_head": [0.000785, -0.001813, -0.002554, -6.1e-05, -0.005508, 0.007999, 0.00953, 0.003282, 0.003827, 0.007871, 0.005057, 0.008517, 0.010833, 0.002242, -0.001557, 0.00092, 0.005365, 0.000772, -0.001816, -0.002576], "action_sum_warmup": 42.124669385484594, "action_sum_measure": 5338.217272675324}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.03347957110963762, "warmup_seconds": 13.355507970089093, "measure_seconds": 85.10608856193721, "boundary_seconds": 0.0005660778842866421}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1175.0040648058161, "init_seconds": 0.03347957110963762, "warmup_seconds": 13.355507970089093, "measure_seconds": 85.10608856193721}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 16, "rollout_calls_measure": 100, "buffer_capacity_rows_per_env": 29000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888, 3622263192, 2939905540]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:03:58.576390+00:00", "extra": {"fingerprint": {"first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669], "actions_head": [0.000779, -0.001996, -0.002624, 7.6e-05, -0.005505, 0.007947, 0.009489, 0.003434, 0.00365, 0.00791, 0.005276, 0.008565, 0.010682, 0.002281, -0.001348, 0.00097, 0.005429, 0.000745, -0.001912, -0.002532], "action_sum_warmup": 42.67921448528796, "action_sum_measure": 5340.831448318613}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.040454025845974684, "warmup_seconds": 13.443762134993449, "measure_seconds": 84.38992843287997, "boundary_seconds": 0.0005403459072113037}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1184.9755279687859, "init_seconds": 0.040454025845974684, "warmup_seconds": 13.443762134993449, "measure_seconds": 84.38992843287997}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 16, "rollout_calls_measure": 100, "buffer_capacity_rows_per_env": 29000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128, 748285535, 1087078637]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:05:36.566842+00:00", "extra": {"fingerprint": {"first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488], "actions_head": [0.000698, -0.002021, -0.00267, -8e-06, -0.005638, 0.007997, 0.009449, 0.003408, 0.003774, 0.00797, 0.005105, 0.008526, 0.01072, 0.002229, -0.001515, 0.000946, 0.005336, 0.000815, -0.001899, -0.002512], "action_sum_warmup": 42.560361873438524, "action_sum_measure": 5330.356120395431}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.028249556897208095, "warmup_seconds": 13.513205276802182, "measure_seconds": 82.95827226387337, "boundary_seconds": 0.0005580680444836617}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1205.425297213524, "init_seconds": 0.028249556897208095, "warmup_seconds": 13.513205276802182, "measure_seconds": 82.95827226387337}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 16, "rollout_calls_measure": 100, "buffer_capacity_rows_per_env": 29000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729, 2938131411, 2102574505]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:07:13.181942+00:00", "extra": {"fingerprint": {"first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156], "actions_head": [0.000719, -0.001974, -0.002608, 5e-06, -0.005607, 0.008015, 0.009431, 0.00335, 0.003707, 0.007888, 0.005197, 0.008656, 0.010754, 0.002205, -0.001483, 0.000932, 0.005332, 0.000732, -0.001744, -0.002633], "action_sum_warmup": 41.95698229321715, "action_sum_measure": 5339.194880631797}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.0259896251372993, "warmup_seconds": 13.224127424880862, "measure_seconds": 83.4097849149257, "boundary_seconds": 0.0005816800985485315}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1198.900106288436, "init_seconds": 0.0259896251372993, "warmup_seconds": 13.224127424880862, "measure_seconds": 83.4097849149257}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 16, "rollout_calls_measure": 100, "buffer_capacity_rows_per_env": 29000, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383, 2051908698, 3189870141]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:08:49.957705+00:00", "extra": {"fingerprint": {"first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815], "actions_head": [0.000727, -0.00187, -0.002521, 0.000104, -0.005546, 0.007989, 0.009465, 0.00333, 0.003745, 0.007968, 0.005126, 0.008495, 0.010742, 0.002223, -0.001424, 0.000974, 0.005404, 0.000728, -0.0019, -0.002511], "action_sum_warmup": 41.90669085659101, "action_sum_measure": 5332.586648690465}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.045133329927921295, "warmup_seconds": 11.329426350072026, "measure_seconds": 71.12101453496143, "boundary_seconds": 0.0009422751609236002}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1406.0541831956339, "init_seconds": 0.045133329927921295, "warmup_seconds": 11.329426350072026, "measure_seconds": 71.12101453496143}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 8, "rollout_calls_measure": 50, "buffer_capacity_rows_per_env": 14500, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933, 2822786858, 3633331186, 2067319448, 3232733299, 2825192541, 3252748741]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:10:12.567404+00:00", "extra": {"fingerprint": {"first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031], "actions_head": [0.000785, -0.001812, -0.002553, -6.1e-05, -0.005506, 0.008001, 0.009531, 0.003281, 0.003828, 0.007871, 0.005055, 0.008518, 0.010835, 0.002243, -0.001557, 0.000921, 0.005368, 0.000773, -0.001817, -0.002576], "action_sum_warmup": 38.98854251801458, "action_sum_measure": 5338.540432940979}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.041277840035036206, "warmup_seconds": 11.289326349971816, "measure_seconds": 71.81625141017139, "boundary_seconds": 0.0009348129387944937}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1392.4424908905357, "init_seconds": 0.041277840035036206, "warmup_seconds": 11.289326349971816, "measure_seconds": 71.81625141017139}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 8, "rollout_calls_measure": 50, "buffer_capacity_rows_per_env": 14500, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888, 3622263192, 2939905540, 3300572547, 629968755, 3493892193, 3289447461]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:11:35.830498+00:00", "extra": {"fingerprint": {"first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669], "actions_head": [0.000778, -0.001997, -0.002623, 7.4e-05, -0.005502, 0.007951, 0.00949, 0.003435, 0.00365, 0.00791, 0.005277, 0.008566, 0.010682, 0.002277, -0.001347, 0.000971, 0.005433, 0.000744, -0.00191, -0.002531], "action_sum_warmup": 40.25147032081986, "action_sum_measure": 5336.460866406052}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.05139086698181927, "warmup_seconds": 11.71073575108312, "measure_seconds": 72.19910867884755, "boundary_seconds": 0.00103938695974648}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1385.0586500287002, "init_seconds": 0.05139086698181927, "warmup_seconds": 11.71073575108312, "measure_seconds": 72.19910867884755}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 8, "rollout_calls_measure": 50, "buffer_capacity_rows_per_env": 14500, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128, 748285535, 1087078637, 2896556764, 2971655972, 501781881, 1396663187]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:12:59.906303+00:00", "extra": {"fingerprint": {"first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488], "actions_head": [0.000698, -0.002021, -0.002672, -1e-05, -0.005635, 0.007999, 0.009449, 0.003406, 0.003773, 0.007971, 0.005106, 0.008527, 0.010722, 0.002229, -0.001515, 0.000947, 0.005338, 0.000815, -0.001901, -0.002512], "action_sum_warmup": 40.159782262224326, "action_sum_measure": 5339.340236489801}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.04268703004345298, "warmup_seconds": 11.392542641842738, "measure_seconds": 71.84602998197079, "boundary_seconds": 0.0010240820702165365}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1391.8653546353812, "init_seconds": 0.04268703004345298, "warmup_seconds": 11.392542641842738, "measure_seconds": 71.84602998197079}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 8, "rollout_calls_measure": 50, "buffer_capacity_rows_per_env": 14500, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729, 2938131411, 2102574505, 2260797203, 3551913963, 3364940778, 1820888441]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:14:23.305760+00:00", "extra": {"fingerprint": {"first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156], "actions_head": [0.000719, -0.001973, -0.002608, 4e-06, -0.005604, 0.008019, 0.009431, 0.00335, 0.003706, 0.00789, 0.005193, 0.008657, 0.010756, 0.002204, -0.001484, 0.000934, 0.005334, 0.00073, -0.001742, -0.002633], "action_sum_warmup": 39.55811711491697, "action_sum_measure": 5337.7119319422945}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.04412218206562102, "warmup_seconds": 11.719555580057204, "measure_seconds": 70.7797764108982, "boundary_seconds": 0.0009097021538764238}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1412.8329456633135, "init_seconds": 0.04412218206562102, "warmup_seconds": 11.719555580057204, "measure_seconds": 70.7797764108982}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 8, "rollout_calls_measure": 50, "buffer_capacity_rows_per_env": 14500, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383, 2051908698, 3189870141, 2134560999, 2845249056, 27589537, 676233087]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:15:45.964465+00:00", "extra": {"fingerprint": {"first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815], "actions_head": [0.000726, -0.00187, -0.002523, 0.000103, -0.005544, 0.007995, 0.009467, 0.003327, 0.003746, 0.007968, 0.005126, 0.008494, 0.010746, 0.002223, -0.001424, 0.000977, 0.005405, 0.000728, -0.001902, -0.002512], "action_sum_warmup": 39.47848598656856, "action_sum_measure": 5338.222295171519}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.06983746797777712, "warmup_seconds": 11.08017460303381, "measure_seconds": 70.26881112297997, "boundary_seconds": 0.002465861849486828}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1423.1064735816635, "init_seconds": 0.06983746797777712, "warmup_seconds": 11.08017460303381, "measure_seconds": 70.26881112297997}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 4, "rollout_calls_measure": 25, "buffer_capacity_rows_per_env": 7250, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933, 2822786858, 3633331186, 2067319448, 3232733299, 2825192541, 3252748741, 3282652602, 2141595418, 1326433267, 3322114159, 899507640, 2393348072, 1585074779, 3540414514]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:17:07.502957+00:00", "extra": {"fingerprint": {"first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031], "actions_head": [0.000785, -0.001812, -0.002553, -6.1e-05, -0.005506, 0.008001, 0.009531, 0.003281, 0.003828, 0.007871, 0.005055, 0.008518, 0.010835, 0.002243, -0.001557, 0.000921, 0.005368, 0.000773, -0.001817, -0.002576], "action_sum_warmup": 37.5880310200846, "action_sum_measure": 5338.816774007841}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.09142868593335152, "warmup_seconds": 11.267568468814716, "measure_seconds": 70.64260354894213, "boundary_seconds": 0.002432740991935134}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1415.576365765153, "init_seconds": 0.09142868593335152, "warmup_seconds": 11.267568468814716, "measure_seconds": 70.64260354894213}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 4, "rollout_calls_measure": 25, "buffer_capacity_rows_per_env": 7250, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888, 3622263192, 2939905540, 3300572547, 629968755, 3493892193, 3289447461, 1818633656, 2310080866, 3463652908, 1379128424, 3135265510, 529413058, 1325280630, 1657318413]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:18:29.619311+00:00", "extra": {"fingerprint": {"first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669], "actions_head": [0.000778, -0.001997, -0.002623, 7.4e-05, -0.005502, 0.007951, 0.00949, 0.003435, 0.00365, 0.00791, 0.005277, 0.008566, 0.010682, 0.002277, -0.001347, 0.000971, 0.005433, 0.000744, -0.00191, -0.002531], "action_sum_warmup": 38.13555700935831, "action_sum_measure": 5335.887483355612}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.07041427213698626, "warmup_seconds": 11.178510640049353, "measure_seconds": 70.84407875803299, "boundary_seconds": 0.0025551868602633476}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1411.5505735002732, "init_seconds": 0.07041427213698626, "warmup_seconds": 11.178510640049353, "measure_seconds": 70.84407875803299}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 4, "rollout_calls_measure": 25, "buffer_capacity_rows_per_env": 7250, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128, 748285535, 1087078637, 2896556764, 2971655972, 501781881, 1396663187, 35535838, 3191457569, 1865027337, 2866184297, 4111885058, 1595357730, 2048696666, 2907168658]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:19:51.829098+00:00", "extra": {"fingerprint": {"first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488], "actions_head": [0.000698, -0.002021, -0.002672, -1e-05, -0.005635, 0.007999, 0.009449, 0.003406, 0.003773, 0.007971, 0.005106, 0.008527, 0.010722, 0.002229, -0.001515, 0.000947, 0.005338, 0.000815, -0.001901, -0.002512], "action_sum_warmup": 38.09703316613013, "action_sum_measure": 5339.729536875882}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.07526857312768698, "warmup_seconds": 11.260136731900275, "measure_seconds": 70.59189216187224, "boundary_seconds": 0.0025613070465624332}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1416.5932791643108, "init_seconds": 0.07526857312768698, "warmup_seconds": 11.260136731900275, "measure_seconds": 70.59189216187224}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 4, "rollout_calls_measure": 25, "buffer_capacity_rows_per_env": 7250, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729, 2938131411, 2102574505, 2260797203, 3551913963, 3364940778, 1820888441, 419499557, 2900963514, 3904039500, 3391033405, 2141692054, 1169319372, 231611715, 3691296346]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:21:13.872686+00:00", "extra": {"fingerprint": {"first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156], "actions_head": [0.000719, -0.001973, -0.002608, 4e-06, -0.005604, 0.008019, 0.009431, 0.00335, 0.003706, 0.00789, 0.005193, 0.008657, 0.010756, 0.002204, -0.001484, 0.000934, 0.005334, 0.00073, -0.001742, -0.002633], "action_sum_warmup": 36.943927240077755, "action_sum_measure": 5331.767515157504}}} +{"schema_version": 1, "framework": "zeroRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.09286610805429518, "warmup_seconds": 11.186236324021593, "measure_seconds": 70.48001181310974, "boundary_seconds": 0.002476111985743046}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1418.8419869333698, "init_seconds": 0.09286610805429518, "warmup_seconds": 11.186236324021593, "measure_seconds": 70.48001181310974}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "rollout_chunk_rows_per_env": 250, "rollout_calls_warmup": 4, "rollout_calls_measure": 25, "buffer_capacity_rows_per_env": 7250, "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383, 2051908698, 3189870141, 2134560999, 2845249056, 27589537, 676233087, 1361455104, 964164021, 2999092853, 24146355, 286365823, 3236068170, 821878104, 1143659019]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:22:35.748276+00:00", "extra": {"fingerprint": {"first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815], "actions_head": [0.000726, -0.00187, -0.002523, 0.000103, -0.005544, 0.007995, 0.009467, 0.003327, 0.003746, 0.007968, 0.005126, 0.008494, 0.010746, 0.002223, -0.001424, 0.000977, 0.005405, 0.000728, -0.001902, -0.002512], "action_sum_warmup": 37.224906347921205, "action_sum_measure": 5330.732297533492}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.019229669123888016, "warmup_seconds": 20.914893965004012, "measure_seconds": 133.85433306894265, "boundary_seconds": 0.00014114007353782654}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 747.0807833205838, "init_seconds": 0.019229669123888016, "warmup_seconds": 20.914893965004012, "measure_seconds": 133.85433306894265}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:25:10.542065+00:00", "extra": {"fingerprint": {"actions_head": [0.000782, -0.001812, -0.002553, -6.2e-05, -0.005506, 0.007999, 0.009535, 0.003281, 0.00383, 0.007872, 0.005056, 0.008516, 0.010834, 0.002244, -0.001553, 0.000921, 0.005364, -0.003047, 0.000427, 0.006886], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.06796349608339369, "warmup_seconds": 21.315753035945818, "measure_seconds": 133.48360912804492, "boundary_seconds": 0.00014356407336890697}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 749.1556502946697, "init_seconds": 0.06796349608339369, "warmup_seconds": 21.315753035945818, "measure_seconds": 133.48360912804492}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:27:45.415348+00:00", "extra": {"fingerprint": {"actions_head": [0.000776, -0.001994, -0.002623, 7.4e-05, -0.005503, 0.007947, 0.009494, 0.003432, 0.003652, 0.00791, 0.005274, 0.008565, 0.010683, 0.002282, -0.001344, 0.00097, 0.005429, -0.001209, -1.5e-05, 0.001186], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.016869707964360714, "warmup_seconds": 20.878526244079694, "measure_seconds": 130.0476731727831, "boundary_seconds": 0.00015293597243726254}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 768.9487828601027, "init_seconds": 0.016869707964360714, "warmup_seconds": 20.878526244079694, "measure_seconds": 130.0476731727831}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:30:16.363793+00:00", "extra": {"fingerprint": {"actions_head": [0.000694, -0.002019, -0.00267, -1e-05, -0.005637, 0.007996, 0.009454, 0.003408, 0.003777, 0.007971, 0.005104, 0.008526, 0.01072, 0.002231, -0.001511, 0.000946, 0.005334, -0.000844, 0.002764, 0.002006], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.018079090863466263, "warmup_seconds": 20.571530971908942, "measure_seconds": 131.9114851260092, "boundary_seconds": 0.00013702898286283016}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 758.0841039313175, "init_seconds": 0.018079090863466263, "warmup_seconds": 20.571530971908942, "measure_seconds": 131.9114851260092}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:32:48.870828+00:00", "extra": {"fingerprint": {"actions_head": [0.000717, -0.001973, -0.002607, 4e-06, -0.005605, 0.008015, 0.009436, 0.00335, 0.003709, 0.007889, 0.005196, 0.008655, 0.010753, 0.002205, -0.001481, 0.000933, 0.005331, 0.001077, 0.004181, 0.000296], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.01807456905953586, "warmup_seconds": 21.234045697841793, "measure_seconds": 135.20165739092045, "boundary_seconds": 0.00015333504416048527}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 739.6359033592406, "init_seconds": 0.01807456905953586, "warmup_seconds": 21.234045697841793, "measure_seconds": 135.20165739092045}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:35:25.330867+00:00", "extra": {"fingerprint": {"actions_head": [0.000724, -0.001868, -0.00252, 0.000102, -0.005545, 0.007989, 0.00947, 0.003329, 0.003748, 0.007969, 0.005124, 0.008495, 0.010744, 0.002225, -0.00142, 0.000975, 0.005403, 0.003445, 0.005474, -0.000733], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.08247828506864607, "warmup_seconds": 15.388123325072229, "measure_seconds": 96.55503996019252, "boundary_seconds": 0.00026456709019839764}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1035.6787179750302, "init_seconds": 0.08247828506864607, "warmup_seconds": 15.388123325072229, "measure_seconds": 96.55503996019252}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:37:17.363097+00:00", "extra": {"fingerprint": {"actions_head": [0.000785, -0.001813, -0.002554, -6.1e-05, -0.005508, 0.007999, 0.00953, 0.003282, 0.003827, 0.007871, 0.005057, 0.008517, 0.010833, 0.002242, -0.001557, 0.00092, 0.005365, 0.000772, -0.001816, -0.002576], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.021479320945218205, "warmup_seconds": 15.553057807963341, "measure_seconds": 97.51021297113039, "boundary_seconds": 0.00024095503613352776}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1025.5336026145976, "init_seconds": 0.021479320945218205, "warmup_seconds": 15.553057807963341, "measure_seconds": 97.51021297113039}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:39:10.453700+00:00", "extra": {"fingerprint": {"actions_head": [0.000779, -0.001996, -0.002624, 7.6e-05, -0.005505, 0.007947, 0.009489, 0.003434, 0.00365, 0.00791, 0.005276, 0.008565, 0.010682, 0.002281, -0.001348, 0.00097, 0.005429, 0.000745, -0.001912, -0.002532], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.0221707159653306, "warmup_seconds": 15.49713539890945, "measure_seconds": 100.1117470501922, "boundary_seconds": 0.00023548398166894913}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 998.8837768445277, "init_seconds": 0.0221707159653306, "warmup_seconds": 15.49713539890945, "measure_seconds": 100.1117470501922}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:41:06.089955+00:00", "extra": {"fingerprint": {"actions_head": [0.000698, -0.002021, -0.00267, -8e-06, -0.005638, 0.007997, 0.009449, 0.003408, 0.003774, 0.00797, 0.005105, 0.008526, 0.01072, 0.002229, -0.001515, 0.000946, 0.005336, 0.000815, -0.001899, -0.002512], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.020086859120056033, "warmup_seconds": 15.788374211173505, "measure_seconds": 97.59652749914676, "boundary_seconds": 0.00023049302399158478}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1024.6266190246804, "init_seconds": 0.020086859120056033, "warmup_seconds": 15.788374211173505, "measure_seconds": 97.59652749914676}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:42:59.500125+00:00", "extra": {"fingerprint": {"actions_head": [0.000719, -0.001974, -0.002608, 5e-06, -0.005607, 0.008015, 0.009431, 0.00335, 0.003707, 0.007888, 0.005197, 0.008656, 0.010754, 0.002205, -0.001483, 0.000932, 0.005332, 0.000732, -0.001744, -0.002633], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.019515922060236335, "warmup_seconds": 15.662469262024388, "measure_seconds": 96.72132846992463, "boundary_seconds": 0.0002145380713045597}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1033.8981234226417, "init_seconds": 0.019515922060236335, "warmup_seconds": 15.662469262024388, "measure_seconds": 96.72132846992463}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:44:51.910038+00:00", "extra": {"fingerprint": {"actions_head": [0.000727, -0.00187, -0.002521, 0.000104, -0.005546, 0.007989, 0.009465, 0.00333, 0.003745, 0.007968, 0.005126, 0.008495, 0.010742, 0.002223, -0.001424, 0.000974, 0.005404, 0.000728, -0.0019, -0.002511], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.0306753390468657, "warmup_seconds": 12.487799840047956, "measure_seconds": 77.7294903320726, "boundary_seconds": 0.00042448495514690876}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1286.5130026298164, "init_seconds": 0.0306753390468657, "warmup_seconds": 12.487799840047956, "measure_seconds": 77.7294903320726}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933, 2822786858, 3633331186]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:46:22.164662+00:00", "extra": {"fingerprint": {"actions_head": [0.000785, -0.001813, -0.002554, -6.1e-05, -0.005508, 0.007999, 0.00953, 0.003282, 0.003827, 0.007871, 0.005057, 0.008517, 0.010833, 0.002242, -0.001557, 0.00092, 0.005365, 0.000772, -0.001816, -0.002576], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.06640250515192747, "warmup_seconds": 12.67382093006745, "measure_seconds": 77.50499380193651, "boundary_seconds": 0.0004256409592926502}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1290.2394425777175, "init_seconds": 0.06640250515192747, "warmup_seconds": 12.67382093006745, "measure_seconds": 77.50499380193651}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888, 3622263192, 2939905540]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:47:52.415606+00:00", "extra": {"fingerprint": {"actions_head": [0.000779, -0.001996, -0.002624, 7.6e-05, -0.005505, 0.007947, 0.009489, 0.003434, 0.00365, 0.00791, 0.005276, 0.008565, 0.010682, 0.002281, -0.001348, 0.00097, 0.005429, 0.000745, -0.001912, -0.002532], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.028010449837893248, "warmup_seconds": 12.3723887600936, "measure_seconds": 78.15528955496848, "boundary_seconds": 0.0004481230862438679}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1279.5039282615364, "init_seconds": 0.028010449837893248, "warmup_seconds": 12.3723887600936, "measure_seconds": 78.15528955496848}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128, 748285535, 1087078637]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:49:22.978284+00:00", "extra": {"fingerprint": {"actions_head": [0.000698, -0.002021, -0.00267, -8e-06, -0.005638, 0.007997, 0.009449, 0.003408, 0.003774, 0.00797, 0.005105, 0.008526, 0.01072, 0.002229, -0.001515, 0.000946, 0.005336, 0.000815, -0.001899, -0.002512], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.02667504409328103, "warmup_seconds": 12.556958184111863, "measure_seconds": 77.45670528500341, "boundary_seconds": 0.0004072990268468857}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1291.043811275578, "init_seconds": 0.02667504409328103, "warmup_seconds": 12.556958184111863, "measure_seconds": 77.45670528500341}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729, 2938131411, 2102574505]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:50:53.025847+00:00", "extra": {"fingerprint": {"actions_head": [0.000719, -0.001974, -0.002608, 5e-06, -0.005607, 0.008015, 0.009431, 0.00335, 0.003707, 0.007888, 0.005197, 0.008656, 0.010754, 0.002205, -0.001483, 0.000932, 0.005332, 0.000732, -0.001744, -0.002633], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.03152474807575345, "warmup_seconds": 12.404628840042278, "measure_seconds": 77.11864354391582, "boundary_seconds": 0.0004119661170989275}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1296.7033055120348, "init_seconds": 0.03152474807575345, "warmup_seconds": 12.404628840042278, "measure_seconds": 77.11864354391582}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383, 2051908698, 3189870141]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:52:22.586728+00:00", "extra": {"fingerprint": {"actions_head": [0.000727, -0.00187, -0.002521, 0.000104, -0.005546, 0.007989, 0.009465, 0.00333, 0.003745, 0.007968, 0.005126, 0.008495, 0.010742, 0.002223, -0.001424, 0.000974, 0.005404, 0.000728, -0.0019, -0.002511], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.04246912500821054, "warmup_seconds": 10.934608245966956, "measure_seconds": 68.68103805207647, "boundary_seconds": 0.0007857279852032661}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1456.0059491846403, "init_seconds": 0.04246912500821054, "warmup_seconds": 10.934608245966956, "measure_seconds": 68.68103805207647}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933, 2822786858, 3633331186, 2067319448, 3232733299, 2825192541, 3252748741]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:53:42.252971+00:00", "extra": {"fingerprint": {"actions_head": [0.000785, -0.001812, -0.002553, -6.1e-05, -0.005506, 0.008001, 0.009531, 0.003281, 0.003828, 0.007871, 0.005055, 0.008518, 0.010835, 0.002243, -0.001557, 0.000921, 0.005368, 0.000773, -0.001817, -0.002576], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.042447553016245365, "warmup_seconds": 11.268678647931665, "measure_seconds": 68.45242487383075, "boundary_seconds": 0.0007522001396864653}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1460.868627872814, "init_seconds": 0.042447553016245365, "warmup_seconds": 11.268678647931665, "measure_seconds": 68.45242487383075}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888, 3622263192, 2939905540, 3300572547, 629968755, 3493892193, 3289447461]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:55:02.024382+00:00", "extra": {"fingerprint": {"actions_head": [0.000778, -0.001997, -0.002623, 7.4e-05, -0.005502, 0.007951, 0.00949, 0.003435, 0.00365, 0.00791, 0.005277, 0.008566, 0.010682, 0.002277, -0.001347, 0.000971, 0.005433, 0.000744, -0.00191, -0.002531], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.0802562958560884, "warmup_seconds": 10.965942787006497, "measure_seconds": 69.69335668697022, "boundary_seconds": 0.0007370628882199526}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1434.856990016322, "init_seconds": 0.0802562958560884, "warmup_seconds": 10.965942787006497, "measure_seconds": 69.69335668697022}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128, 748285535, 1087078637, 2896556764, 2971655972, 501781881, 1396663187]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:56:22.772620+00:00", "extra": {"fingerprint": {"actions_head": [0.000698, -0.002021, -0.002672, -1e-05, -0.005635, 0.007999, 0.009449, 0.003406, 0.003773, 0.007971, 0.005106, 0.008527, 0.010722, 0.002229, -0.001515, 0.000947, 0.005338, 0.000815, -0.001901, -0.002512], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.04370272206142545, "warmup_seconds": 11.2733521850314, "measure_seconds": 69.41808202513494, "boundary_seconds": 0.0007938230410218239}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1440.5468587246755, "init_seconds": 0.04370272206142545, "warmup_seconds": 11.2733521850314, "measure_seconds": 69.41808202513494}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729, 2938131411, 2102574505, 2260797203, 3551913963, 3364940778, 1820888441]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:57:43.515770+00:00", "extra": {"fingerprint": {"actions_head": [0.000719, -0.001973, -0.002608, 4e-06, -0.005604, 0.008019, 0.009431, 0.00335, 0.003706, 0.00789, 0.005193, 0.008657, 0.010756, 0.002204, -0.001484, 0.000934, 0.005334, 0.00073, -0.001742, -0.002633], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.046870254911482334, "warmup_seconds": 10.970491142012179, "measure_seconds": 68.10395700111985, "boundary_seconds": 0.0007612018380314112}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1468.3434620158073, "init_seconds": 0.046870254911482334, "warmup_seconds": 10.970491142012179, "measure_seconds": 68.10395700111985}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383, 2051908698, 3189870141, 2134560999, 2845249056, 27589537, 676233087]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-04T23:59:02.645730+00:00", "extra": {"fingerprint": {"actions_head": [0.000726, -0.00187, -0.002523, 0.000103, -0.005544, 0.007995, 0.009467, 0.003327, 0.003746, 0.007968, 0.005126, 0.008494, 0.010746, 0.002223, -0.001424, 0.000977, 0.005405, 0.000728, -0.001902, -0.002512], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.0951613369397819, "warmup_seconds": 10.8892914110329, "measure_seconds": 68.83251529280096, "boundary_seconds": 0.0014585109893232584}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1452.80176926004, "init_seconds": 0.0951613369397819, "warmup_seconds": 10.8892914110329, "measure_seconds": 68.83251529280096}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933, 2822786858, 3633331186, 2067319448, 3232733299, 2825192541, 3252748741, 3282652602, 2141595418, 1326433267, 3322114159, 899507640, 2393348072, 1585074779, 3540414514]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:00:22.473811+00:00", "extra": {"fingerprint": {"actions_head": [0.000785, -0.001812, -0.002553, -6.1e-05, -0.005506, 0.008001, 0.009531, 0.003281, 0.003828, 0.007871, 0.005055, 0.008518, 0.010835, 0.002243, -0.001557, 0.000921, 0.005368, 0.000773, -0.001817, -0.002576], "action_sum_warmup": 1, "action_sum_measure": 1, "first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.08256699307821691, "warmup_seconds": 10.96045124786906, "measure_seconds": 68.77023433987051, "boundary_seconds": 0.0013800389133393764}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1454.117482074997, "init_seconds": 0.08256699307821691, "warmup_seconds": 10.96045124786906, "measure_seconds": 68.77023433987051}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888, 3622263192, 2939905540, 3300572547, 629968755, 3493892193, 3289447461, 1818633656, 2310080866, 3463652908, 1379128424, 3135265510, 529413058, 1325280630, 1657318413]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:01:42.297662+00:00", "extra": {"fingerprint": {"actions_head": [0.000778, -0.001997, -0.002623, 7.4e-05, -0.005502, 0.007951, 0.00949, 0.003435, 0.00365, 0.00791, 0.005277, 0.008566, 0.010682, 0.002277, -0.001347, 0.000971, 0.005433, 0.000744, -0.00191, -0.002531], "action_sum_warmup": 1, "action_sum_measure": 3, "first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.07249839208088815, "warmup_seconds": 11.015183890936896, "measure_seconds": 69.46931593911722, "boundary_seconds": 0.0014718370512127876}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1439.4844493306916, "init_seconds": 0.07249839208088815, "warmup_seconds": 11.015183890936896, "measure_seconds": 69.46931593911722}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128, 748285535, 1087078637, 2896556764, 2971655972, 501781881, 1396663187, 35535838, 3191457569, 1865027337, 2866184297, 4111885058, 1595357730, 2048696666, 2907168658]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:03:02.866063+00:00", "extra": {"fingerprint": {"actions_head": [0.000698, -0.002021, -0.002672, -1e-05, -0.005635, 0.007999, 0.009449, 0.003406, 0.003773, 0.007971, 0.005106, 0.008527, 0.010722, 0.002229, -0.001515, 0.000947, 0.005338, 0.000815, -0.001901, -0.002512], "action_sum_warmup": 0, "action_sum_measure": 2, "first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.10201140004210174, "warmup_seconds": 11.07774443202652, "measure_seconds": 69.26369711989537, "boundary_seconds": 0.0014257729053497314}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1443.7577570671708, "init_seconds": 0.10201140004210174, "warmup_seconds": 11.07774443202652, "measure_seconds": 69.26369711989537}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729, 2938131411, 2102574505, 2260797203, 3551913963, 3364940778, 1820888441, 419499557, 2900963514, 3904039500, 3391033405, 2141692054, 1169319372, 231611715, 3691296346]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:04:23.320043+00:00", "extra": {"fingerprint": {"actions_head": [0.000719, -0.001973, -0.002608, 4e-06, -0.005604, 0.008019, 0.009431, 0.00335, 0.003706, 0.00789, 0.005193, 0.008657, 0.010756, 0.002204, -0.001484, 0.000934, 0.005334, 0.00073, -0.001742, -0.002633], "action_sum_warmup": 0, "action_sum_measure": 0, "first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156]}}} +{"schema_version": 1, "framework": "SB3", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.07487382716499269, "warmup_seconds": 11.062920209020376, "measure_seconds": 69.11567407706752, "boundary_seconds": 0.0015138292219489813}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1446.8498113538596, "init_seconds": 0.07487382716499269, "warmup_seconds": 11.062920209020376, "measure_seconds": 69.11567407706752}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "vec_env": "DummyVecEnv", "collection": "SharedTrunkPolicy.forward_collection (action, value, log_prob, entropy), deterministic", "storage": "RolloutBuffer + entropy side array", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383, 2051908698, 3189870141, 2134560999, 2845249056, 27589537, 676233087, 1361455104, 964164021, 2999092853, 24146355, 286365823, 3236068170, 821878104, 1143659019]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:05:43.583678+00:00", "extra": {"fingerprint": {"actions_head": [0.000726, -0.00187, -0.002523, 0.000103, -0.005544, 0.007995, 0.009467, 0.003327, 0.003746, 0.007968, 0.005126, 0.008494, 0.010746, 0.002223, -0.001424, 0.000977, 0.005405, 0.000728, -0.001902, -0.002512], "action_sum_warmup": 0, "action_sum_measure": 1, "first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.24264180101454258, "warmup_seconds": 43.268152808072045, "measure_seconds": 264.32697346503846, "boundary_seconds": 0.009441740810871124}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 378.31931675042097, "init_seconds": 0.24264180101454258, "warmup_seconds": 43.268152808072045, "measure_seconds": 264.32697346503846}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 6250, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:10:51.827467+00:00", "extra": {"fingerprint": {"actions_head": [0.000782, -0.003047, -0.006523, -0.000289, 0.000503, -0.010871, -0.011366, -0.004242, -0.004122, -0.004104, 0.001986, -0.005265, -0.00462, -0.008231, -0.011135, -0.005053, -0.004814, -0.002592, -0.003196, -0.001817], "action_sum_warmup": 744, "action_sum_measure": 4631, "first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.04270713892765343, "warmup_seconds": 42.4182735059876, "measure_seconds": 264.7489486981649, "boundary_seconds": 0.009210046846419573}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 377.7163251892949, "init_seconds": 0.04270713892765343, "warmup_seconds": 42.4182735059876, "measure_seconds": 264.7489486981649}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 6250, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:15:59.050486+00:00", "extra": {"fingerprint": {"actions_head": [0.000776, -0.001209, -0.000451, 8e-06, 0.001824, 0.002345, -0.010423, -0.003787, -0.003489, -0.003021, -0.004278, -0.008357, -0.001517, -0.004337, -0.00536, -0.005012, -0.00265, -0.002265, -0.001827, -0.00178], "action_sum_warmup": 744, "action_sum_measure": 4629, "first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.044441396836191416, "warmup_seconds": 42.29434553300962, "measure_seconds": 263.7600160881411, "boundary_seconds": 0.010089030023664236}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 379.1325216123085, "init_seconds": 0.044441396836191416, "warmup_seconds": 42.29434553300962, "measure_seconds": 263.7600160881411}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 6250, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:21:05.163329+00:00", "extra": {"fingerprint": {"actions_head": [0.000694, -0.000844, -0.001726, -0.000612, -0.009453, -0.006204, -0.011679, -0.008757, -0.005015, -0.004925, -0.00612, -0.009889, -0.00401, -0.007841, -0.005271, -0.005489, -0.007646, -0.006593, -0.005444, -0.003198], "action_sum_warmup": 739, "action_sum_measure": 4624, "first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.04169046203605831, "warmup_seconds": 42.09071990894154, "measure_seconds": 263.4027525759302, "boundary_seconds": 0.009026092011481524}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 379.646753961591, "init_seconds": 0.04169046203605831, "warmup_seconds": 42.09071990894154, "measure_seconds": 263.4027525759302}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 6250, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:26:10.711645+00:00", "extra": {"fingerprint": {"actions_head": [0.000717, 0.001077, -0.000907, -0.001347, 0.001266, 0.001853, -0.010759, -0.002955, -0.005813, -0.005411, -0.004787, -0.002229, -0.004497, -0.005833, -0.002044, -0.007366, -0.006678, -0.00433, -0.00258, -0.002026], "action_sum_warmup": 739, "action_sum_measure": 4636, "first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 1, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.042206825921311975, "warmup_seconds": 42.53437147382647, "measure_seconds": 264.6742093870416, "boundary_seconds": 0.00942200399003923}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 377.8229855927019, "init_seconds": 0.042206825921311975, "warmup_seconds": 42.53437147382647, "measure_seconds": 264.6742093870416}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 6250, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:31:17.975749+00:00", "extra": {"fingerprint": {"actions_head": [0.000724, 0.003445, 0.003415, 9.9e-05, 0.001725, 8e-06, -0.015488, -0.011055, -0.003705, -0.003194, -0.007415, 0.001419, -0.004122, -0.004738, -0.005638, -0.005058, -0.002253, -0.002179, -0.00182, -0.001708], "action_sum_warmup": 739, "action_sum_measure": 4629, "first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.05090545699931681, "warmup_seconds": 26.5357569581829, "measure_seconds": 164.24753502500243, "boundary_seconds": 0.009419663110747933}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 608.8371431862133, "init_seconds": 0.05090545699931681, "warmup_seconds": 26.5357569581829, "measure_seconds": 164.24753502500243}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 6250, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:34:28.823438+00:00", "extra": {"fingerprint": {"actions_head": [0.000785, -0.003042, -0.006522, -0.000284, 0.000505, -0.010835, -0.011373, -0.004257, -0.004123, -0.004106, 0.001986, -0.002643, -0.004348, 0.001387, -0.005322, -0.005537, -0.004859, -0.003182, -0.003104, -0.001819], "action_sum_warmup": 741, "action_sum_measure": 4635, "first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.05007715802639723, "warmup_seconds": 25.997098362073302, "measure_seconds": 164.92196902399883, "boundary_seconds": 0.009415697073563933}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 606.3473568245379, "init_seconds": 0.05007715802639723, "warmup_seconds": 25.997098362073302, "measure_seconds": 164.92196902399883}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 6250, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:37:39.806366+00:00", "extra": {"fingerprint": {"actions_head": [0.000779, -0.001206, -0.000447, 1e-05, 0.001826, 0.002345, -0.010423, -0.003786, -0.003492, -0.003022, -0.004272, -0.008359, -0.001515, -0.004331, -0.00536, -0.005013, -0.002649, -0.002261, -0.001828, -0.001783], "action_sum_warmup": 743, "action_sum_measure": 4632, "first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.05118784215301275, "warmup_seconds": 25.93194957310334, "measure_seconds": 163.67908357200213, "boundary_seconds": 0.009571735048666596}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 610.951612250506, "init_seconds": 0.05118784215301275, "warmup_seconds": 25.93194957310334, "measure_seconds": 163.67908357200213}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 6250, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:40:49.482584+00:00", "extra": {"fingerprint": {"actions_head": [0.000698, -0.000844, -0.001725, -0.00061, -0.009453, -0.006203, -0.011683, -0.008759, -0.005016, -0.004925, -0.006118, -0.009888, -0.004008, -0.00784, -0.005263, -0.007636, -0.009816, -0.005323, -0.005363, -0.003236], "action_sum_warmup": 741, "action_sum_measure": 4626, "first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.05421289894729853, "warmup_seconds": 26.795753760961816, "measure_seconds": 174.28842685185373, "boundary_seconds": 0.012000509072095156}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 573.7615618333663, "init_seconds": 0.05421289894729853, "warmup_seconds": 26.795753760961816, "measure_seconds": 174.28842685185373}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 6250, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:44:10.636554+00:00", "extra": {"fingerprint": {"actions_head": [0.000719, 0.00109, -0.000903, -0.001345, 0.001268, 0.001875, -0.01076, -0.002961, -0.005811, -0.005399, -0.004776, -0.002231, -0.004497, -0.005838, -0.002048, -0.007366, -0.006678, -0.004326, -0.002588, -0.002023], "action_sum_warmup": 741, "action_sum_measure": 4629, "first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 2, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.05510842986404896, "warmup_seconds": 27.97829413600266, "measure_seconds": 174.2710230059456, "boundary_seconds": 0.010587521130219102}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 573.8188614213178, "init_seconds": 0.05510842986404896, "warmup_seconds": 27.97829413600266, "measure_seconds": 174.2710230059456}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 6250, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:47:32.955517+00:00", "extra": {"fingerprint": {"actions_head": [0.000727, 0.003448, 0.003415, 0.0001, 0.001728, 1e-05, -0.015488, -0.011047, -0.003708, -0.003195, -0.007413, 0.001416, -0.004122, -0.004736, -0.005639, -0.005059, -0.002257, -0.002182, -0.001821, -0.001709], "action_sum_warmup": 740, "action_sum_measure": 4628, "first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.06946765491738915, "warmup_seconds": 18.02997309085913, "measure_seconds": 116.81572390696965, "boundary_seconds": 0.010090921074151993}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 856.0491400938332, "init_seconds": 0.06946765491738915, "warmup_seconds": 18.02997309085913, "measure_seconds": 116.81572390696965}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 5000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933, 2822786858, 3633331186]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:49:47.885644+00:00", "extra": {"fingerprint": {"actions_head": [0.000785, -0.003042, -0.006522, -0.000284, 0.000505, -0.010835, -0.011373, -0.004257, -0.004123, -0.004106, 0.001986, -0.002643, -0.004348, 0.001387, -0.005322, -0.005537, -0.004859, -0.003182, -0.003104, -0.001819], "action_sum_warmup": 743, "action_sum_measure": 4633, "first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.07192098582163453, "warmup_seconds": 18.569342420902103, "measure_seconds": 117.29630551300943, "boundary_seconds": 0.010474126087501645}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 852.5417707116864, "init_seconds": 0.07192098582163453, "warmup_seconds": 18.569342420902103, "measure_seconds": 117.29630551300943}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 5000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888, 3622263192, 2939905540]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:52:03.838532+00:00", "extra": {"fingerprint": {"actions_head": [0.000779, -0.001206, -0.000447, 1e-05, 0.001826, 0.002345, -0.010423, -0.003786, -0.003492, -0.003022, -0.004272, -0.008359, -0.001515, -0.004331, -0.00536, -0.005013, -0.002649, -0.002261, -0.001828, -0.001783], "action_sum_warmup": 739, "action_sum_measure": 4634, "first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.07249316899105906, "warmup_seconds": 18.84889970207587, "measure_seconds": 117.1561826649122, "boundary_seconds": 0.010796849150210619}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 853.5614401675927, "init_seconds": 0.07249316899105906, "warmup_seconds": 18.84889970207587, "measure_seconds": 117.1561826649122}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 5000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128, 748285535, 1087078637]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:54:19.930650+00:00", "extra": {"fingerprint": {"actions_head": [0.000698, -0.000844, -0.001725, -0.00061, -0.009453, -0.006203, -0.011683, -0.008759, -0.005016, -0.004925, -0.006118, -0.009888, -0.004008, -0.00784, -0.005263, -0.007636, -0.009816, -0.005323, -0.005363, -0.003236], "action_sum_warmup": 740, "action_sum_measure": 4622, "first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.06654245080426335, "warmup_seconds": 17.956858031917363, "measure_seconds": 111.54461668385193, "boundary_seconds": 0.009422172093763947}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 896.5022515019928, "init_seconds": 0.06654245080426335, "warmup_seconds": 17.956858031917363, "measure_seconds": 111.54461668385193}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 5000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729, 2938131411, 2102574505]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:56:29.512430+00:00", "extra": {"fingerprint": {"actions_head": [0.000719, 0.00109, -0.000903, -0.001345, 0.001268, 0.001875, -0.01076, -0.002961, -0.005811, -0.005399, -0.004776, -0.002231, -0.004497, -0.005838, -0.002048, -0.007366, -0.006678, -0.004326, -0.002588, -0.002023], "action_sum_warmup": 743, "action_sum_measure": 4635, "first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 4, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.0662951641716063, "warmup_seconds": 18.50983667303808, "measure_seconds": 118.63964449591003, "boundary_seconds": 0.010534717002883554}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 842.8885675179799, "init_seconds": 0.0662951641716063, "warmup_seconds": 18.50983667303808, "measure_seconds": 118.63964449591003}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 5000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383, 2051908698, 3189870141]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T00:58:46.744349+00:00", "extra": {"fingerprint": {"actions_head": [0.000727, 0.003448, 0.003415, 0.0001, 0.001728, 1e-05, -0.015488, -0.011047, -0.003708, -0.003195, -0.007413, 0.001416, -0.004122, -0.004736, -0.005639, -0.005059, -0.002257, -0.002182, -0.001821, -0.001709], "action_sum_warmup": 738, "action_sum_measure": 4633, "first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.10428986605256796, "warmup_seconds": 14.331693889806047, "measure_seconds": 90.1761459580157, "boundary_seconds": 0.010660642990842462}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1108.940717499261, "init_seconds": 0.10428986605256796, "warmup_seconds": 14.331693889806047, "measure_seconds": 90.1761459580157}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 5000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933, 2822786858, 3633331186, 2067319448, 3232733299, 2825192541, 3252748741]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T01:00:31.370680+00:00", "extra": {"fingerprint": {"actions_head": [0.000785, -0.00304, -0.006523, -0.000284, 0.000506, -0.010833, -0.011373, -0.004257, -0.004123, -0.004107, 0.001985, -0.00265, -0.00435, 0.005764, -0.004168, -0.00576, -0.0046, -0.002864, -0.003347, -0.001852], "action_sum_warmup": 742, "action_sum_measure": 4634, "first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.10497704287990928, "warmup_seconds": 14.593682903097942, "measure_seconds": 90.31369132292457, "boundary_seconds": 0.010685187997296453}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1107.2518300956294, "init_seconds": 0.10497704287990928, "warmup_seconds": 14.593682903097942, "measure_seconds": 90.31369132292457}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 5000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888, 3622263192, 2939905540, 3300572547, 629968755, 3493892193, 3289447461]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T01:02:16.397256+00:00", "extra": {"fingerprint": {"actions_head": [0.000778, -0.001205, -0.000445, 1.2e-05, 0.001827, 0.002343, -0.01042, -0.003785, -0.003493, -0.003023, -0.004274, -0.008356, -0.001516, -0.004333, -0.005363, -0.005014, -0.00265, -0.002267, -0.001827, -0.001782], "action_sum_warmup": 740, "action_sum_measure": 4631, "first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.09749194094911218, "warmup_seconds": 13.823926555924118, "measure_seconds": 86.28843245701864, "boundary_seconds": 0.010140500031411648}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1158.9038895776819, "init_seconds": 0.09749194094911218, "warmup_seconds": 13.823926555924118, "measure_seconds": 86.28843245701864}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 5000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128, 748285535, 1087078637, 2896556764, 2971655972, 501781881, 1396663187]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T01:03:56.621739+00:00", "extra": {"fingerprint": {"actions_head": [0.000698, -0.000843, -0.001723, -0.00061, -0.009452, -0.006202, -0.011683, -0.008758, -0.005017, -0.004927, -0.00612, -0.009887, -0.004007, -0.007841, -0.005265, -0.007635, -0.009816, -0.00535, -0.005366, -0.003238], "action_sum_warmup": 743, "action_sum_measure": 4622, "first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.09266588906757534, "warmup_seconds": 14.396749258972704, "measure_seconds": 87.84618636011146, "boundary_seconds": 0.01083960197865963}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1138.3533439922585, "init_seconds": 0.09266588906757534, "warmup_seconds": 14.396749258972704, "measure_seconds": 87.84618636011146}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 5000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729, 2938131411, 2102574505, 2260797203, 3551913963, 3364940778, 1820888441]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T01:05:38.972743+00:00", "extra": {"fingerprint": {"actions_head": [0.000719, 0.001091, -0.000901, -0.001343, 0.001272, 0.001875, -0.010759, -0.002962, -0.005813, -0.005405, -0.004779, -0.002232, -0.00449, -0.005835, -0.002051, -0.00737, -0.00668, -0.004333, -0.00259, -0.002025], "action_sum_warmup": 744, "action_sum_measure": 4631, "first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 8, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.10529707581736147, "warmup_seconds": 13.815331602934748, "measure_seconds": 89.29148520715535, "boundary_seconds": 0.010151442838832736}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1119.9276142400477, "init_seconds": 0.10529707581736147, "warmup_seconds": 13.815331602934748, "measure_seconds": 89.29148520715535}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 5000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383, 2051908698, 3189870141, 2134560999, 2845249056, 27589537, 676233087]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T01:07:22.199297+00:00", "extra": {"fingerprint": {"actions_head": [0.000726, 0.003449, 0.003416, 0.0001, 0.00173, 1e-05, -0.015488, -0.011047, -0.003708, -0.003197, -0.007411, 0.001416, -0.004124, -0.004737, -0.005645, -0.00506, -0.002258, -0.002181, -0.001818, -0.001711], "action_sum_warmup": 739, "action_sum_measure": 4620, "first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 0, "seed": 1000, "status": "ok", "phases": {"init_seconds": 0.17019658908247948, "warmup_seconds": 12.315249501960352, "measure_seconds": 74.68386108404957, "boundary_seconds": 0.012448681052774191}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1338.9773713956693, "init_seconds": 0.17019658908247948, "warmup_seconds": 12.315249501960352, "measure_seconds": 74.68386108404957}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 4000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1484808508, 3739256933, 2822786858, 3633331186, 2067319448, 3232733299, 2825192541, 3252748741, 3282652602, 2141595418, 1326433267, 3322114159, 899507640, 2393348072, 1585074779, 3540414514]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T01:08:49.385738+00:00", "extra": {"fingerprint": {"actions_head": [0.000785, -0.00304, -0.006523, -0.000284, 0.000506, -0.010833, -0.011373, -0.004257, -0.004123, -0.004107, 0.001985, -0.00265, -0.00435, 0.005764, -0.004168, -0.00576, -0.0046, -0.002864, -0.003347, -0.001852], "action_sum_warmup": 743, "action_sum_measure": 4624, "first_obs_head": [0.114739, 1.00395, -0.004942, 0.008031]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 1, "seed": 1001, "status": "ok", "phases": {"init_seconds": 0.16921710199676454, "warmup_seconds": 11.92181992996484, "measure_seconds": 74.03750044410117, "boundary_seconds": 0.012206300860270858}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1350.666883676073, "init_seconds": 0.16921710199676454, "warmup_seconds": 11.92181992996484, "measure_seconds": 74.03750044410117}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 4000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1909074423, 3962977888, 3622263192, 2939905540, 3300572547, 629968755, 3493892193, 3289447461, 1818633656, 2310080866, 3463652908, 1379128424, 3135265510, 529413058, 1325280630, 1657318413]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T01:10:15.530874+00:00", "extra": {"fingerprint": {"actions_head": [0.000778, -0.001205, -0.000445, 1.2e-05, 0.001827, 0.002343, -0.01042, -0.003785, -0.003493, -0.003023, -0.004274, -0.008356, -0.001516, -0.004333, -0.005363, -0.005014, -0.00265, -0.002267, -0.001827, -0.001782], "action_sum_warmup": 742, "action_sum_measure": 4623, "first_obs_head": [0.10194, 1.005352, -0.002477, -0.006669]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 2, "seed": 1002, "status": "ok", "phases": {"init_seconds": 0.15892536100000143, "warmup_seconds": 11.927118697902188, "measure_seconds": 74.2007152249571, "boundary_seconds": 0.012118977028876543}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1347.6959042352387, "init_seconds": 0.15892536100000143, "warmup_seconds": 11.927118697902188, "measure_seconds": 74.2007152249571}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 4000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1780739998, 1935968128, 748285535, 1087078637, 2896556764, 2971655972, 501781881, 1396663187, 35535838, 3191457569, 1865027337, 2866184297, 4111885058, 1595357730, 2048696666, 2907168658]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T01:11:41.834122+00:00", "extra": {"fingerprint": {"actions_head": [0.000698, -0.000843, -0.001723, -0.00061, -0.009452, -0.006202, -0.011683, -0.008758, -0.005017, -0.004927, -0.00612, -0.009887, -0.004007, -0.007841, -0.005265, -0.007635, -0.009816, -0.00535, -0.005366, -0.003238], "action_sum_warmup": 743, "action_sum_measure": 4625, "first_obs_head": [0.0995, 0.990195, 0.005593, 0.004488]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 3, "seed": 1003, "status": "ok", "phases": {"init_seconds": 0.16301899612881243, "warmup_seconds": 11.772763726999983, "measure_seconds": 74.3182741838973, "boundary_seconds": 0.013626551954075694}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1345.5640769126903, "init_seconds": 0.16301899612881243, "warmup_seconds": 11.772763726999983, "measure_seconds": 74.3182741838973}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 4000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [3534947301, 1211106729, 2938131411, 2102574505, 2260797203, 3551913963, 3364940778, 1820888441, 419499557, 2900963514, 3904039500, 3391033405, 2141692054, 1169319372, 231611715, 3691296346]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T01:13:08.106326+00:00", "extra": {"fingerprint": {"actions_head": [0.000719, 0.001091, -0.000901, -0.001343, 0.001272, 0.001875, -0.010759, -0.002962, -0.005813, -0.005405, -0.004779, -0.002232, -0.00449, -0.005835, -0.002051, -0.00737, -0.00668, -0.004333, -0.00259, -0.002025], "action_sum_warmup": 740, "action_sum_measure": 4617, "first_obs_head": [0.103652, 0.996273, -0.001525, -0.001156]}}} +{"schema_version": 1, "framework": "TorchRL", "experiment": "A", "mode": "sync", "num_envs": 16, "num_workers": 1, "protocol": "collection", "rep": 4, "seed": 1004, "status": "ok", "phases": {"init_seconds": 0.162312334170565, "warmup_seconds": 12.021811238955706, "measure_seconds": 75.72016698890366, "boundary_seconds": 0.012672435957938433}, "transitions": {"warmup_target": 16000, "warmup_actual": 16000, "measure_target": 100000, "measured": 100000, "exact": true}, "metrics": {"throughput_mean": 1320.6521324055507, "init_seconds": 0.162312334170565, "warmup_seconds": 12.021811238955706, "measure_seconds": 75.72016698890366}, "config": {"device": "cuda", "env_id": "HumanoidStandup-v5", "warmup": 16000, "measure": 100000, "reference_seed": 4242, "collector": "Collector(backend='direct')", "frames_per_batch_warmup": 8000, "frames_per_batch_measure": 4000, "autoreset": "SyncVectorEnv SAME_STEP", "fields": {"action": 100000, "log_prob": 100000, "entropy": 100000, "value": 100000}, "seeds": [1458279434, 2094128383, 2051908698, 3189870141, 2134560999, 2845249056, 27589537, 676233087, 1361455104, 964164021, 2999092853, 24146355, 286365823, 3236068170, 821878104, 1143659019]}, "error": null, "environment": {"python": "3.13.14", "platform": "Linux-6.8.12-tegra-aarch64-with-glibc2.39", "processor": "aarch64", "device": "cuda", "hostname": "gpu1", "torch": "2.13.0+cu130", "torch_num_threads": 14, "cuda_available": true, "gymnasium": "1.3.0", "sb3": "2.9.0", "torchrl": "0.14.1", "tensordict": "0.14.3", "git_sha": "7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4"}, "timestamp_utc": "2026-11-05T01:14:36.027310+00:00", "extra": {"fingerprint": {"actions_head": [0.000726, 0.003449, 0.003416, 0.0001, 0.00173, 1e-05, -0.015488, -0.011047, -0.003708, -0.003197, -0.007411, 0.001416, -0.004124, -0.004737, -0.005645, -0.00506, -0.002258, -0.002181, -0.001818, -0.001711], "action_sum_warmup": 739, "action_sum_measure": 4616, "first_obs_head": [0.114366, 0.990608, 0.007721, -0.000815]}}} diff --git a/benchmarks/resources/plots.py b/benchmarks/resources/plots.py index 799c047..ae97d6f 100644 --- a/benchmarks/resources/plots.py +++ b/benchmarks/resources/plots.py @@ -228,7 +228,7 @@ def _plot_resource_file(res_file: Path, out_dir: Path, fws = [fw for fw in FRAMEWORK_ORDER if any(fw in cells[w] for w in workloads)] paths: list[Path] = [] - workload_device = "cuda" if suffix else "cpu" + workload_device = ("cuda" if suffix.endswith("_gpu") else "cpu") def _vals(wl: str, fw: str, getter): return [v for r in cells[wl].get(fw, []) @@ -302,8 +302,9 @@ def _grouped(ax, getter, ylabel: str, title: str, *, scale: float = 1.0, # 4) GPU telemetry (only when available in THIS side's rows) gpu_rows = [r for r in rows if r["extra"]["resources"].get("gpu", {}).get("available")] - telemetry_name = (f"resources_gpu{suffix}.png" if not suffix - else f"resources_telemetry{suffix}.png") + telemetry_name = (f"resources_telemetry{suffix}.png" + if suffix.endswith("_gpu") + else f"resources_gpu{suffix}.png") if gpu_rows: fig, axes = plt.subplots(1, 2, figsize=(11, 4)) g = lambda r, k: r["extra"]["resources"]["gpu"].get(k) # noqa: E731 @@ -368,21 +369,25 @@ def _grouped(ax, getter, ylabel: str, title: str, *, scale: float = 1.0, def plot_resources(results_dir: Path, out_dir: Path, - suffix: str = "") -> list[Path]: + suffix: str = "", *, slug: str = "") -> list[Path]: """Resource-consumption figures (experiment R), both device sides. - Renders ``resources{suffix}.jsonl`` (CPU side with the default - ``suffix=""``; existing figure names unchanged), then — only when no - explicit suffix is given — auto-detects ``resources_gpu.jsonl`` and - renders the GPU side with ``suffix="_gpu"``. Statistics/telemetry - availability are always evaluated within one side, never combined. + Renders ``resources{slug}.jsonl`` (CPU side; default ``slug=""`` keeps + existing figure names byte-identical), then — only when no explicit + suffix is given — auto-detects ``resources{slug}_gpu.jsonl`` and + renders the GPU side with ``{slug}_gpu`` figure suffix. Statistics/ + telemetry availability are always evaluated within one side, never + combined. """ - paths = _plot_resource_file(results_dir / f"resources{suffix}.jsonl", - out_dir, suffix) + sfx = f"_{slug}" if slug else "" + cpu_sfx = f"{sfx}{suffix}" + paths = _plot_resource_file(results_dir / f"resources{cpu_sfx}.jsonl", + out_dir, cpu_sfx) if not suffix: - gpu_file = results_dir / "resources_gpu.jsonl" + gpu_sfx = f"{sfx}_gpu" + gpu_file = results_dir / f"resources{gpu_sfx}.jsonl" if gpu_file.exists(): - paths += _plot_resource_file(gpu_file, out_dir, "_gpu") + paths += _plot_resource_file(gpu_file, out_dir, gpu_sfx) return paths @@ -392,6 +397,9 @@ def build_parser() -> argparse.ArgumentParser: default=str(REPO_ROOT / "benchmarks/resources/results")) p.add_argument("--out", default=None, help="figure output dir (default: /figures)") + p.add_argument("--env-slug", default="", + help="env slug used in result/figure filenames " + "(default '': reads resources.jsonl, unchanged)") return p @@ -399,7 +407,7 @@ def main(argv: list[str] | None = None) -> int: args = build_parser().parse_args(argv) results_dir = Path(args.results_dir) out_dir = Path(args.out) if args.out else results_dir / "figures" - paths = plot_resources(results_dir, out_dir) + paths = plot_resources(results_dir, out_dir, slug=args.env_slug) for p in paths: print(f"[fig] {p}") print(f"generated {len(paths)} figures") diff --git a/benchmarks/resources/report.py b/benchmarks/resources/report.py index 17d597b..4ef43c0 100644 --- a/benchmarks/resources/report.py +++ b/benchmarks/resources/report.py @@ -146,17 +146,19 @@ def _resources_table(ok: list[dict]) -> str: return _table(header, body) -def _section_resources(results_dir: Path) -> list[str]: +def _section_resources(results_dir: Path, slug: str = "") -> list[str]: """Table + telemetry notes for the resource-consumption campaign (R).""" from benchmarks.scalability.results_schema import load_rows - res_file = results_dir / "resources.jsonl" + sfx = f"_{slug}" if slug else "" + tag = f" [{slug}]" if slug else "" + res_file = results_dir / f"resources{sfx}.jsonl" if not res_file.exists(): return [] rows_all = load_rows(res_file) ok = [r for r in rows_all if r["status"] == "ok"] out = [ - "## Resource consumption (experiment R)", "", + "## Resource consumption (experiment R)" + tag, "", "Fixed workloads, device CPU. CPU% = sum over the live process " "tree (multiprocess sums may exceed 100%); RSS summed across " "tree processes (shared pages may repeat — PSS/USS recorded in " @@ -176,24 +178,26 @@ def _section_resources(results_dir: Path) -> list[str]: out += [ f"- GPU telemetry: {'available (device-level only)' if gpu.get('available') else 'unavailable — ' + str(gpu.get('reason', ''))}.", f"- Energy: {'available — ' + str(energy.get('scope', '')) if energy.get('available') else 'unavailable — ' + str(energy.get('reason', ''))}.", - "- Raw per-run rows: `results/resources.jsonl` (validation runs " + f"- Raw per-run rows: `results/resources{sfx}.jsonl` (validation runs " "kept separately in `resources_validation.*`).", "", ] out += _memory_subsections(ok) return out -def _section_resources_gpu(results_dir: Path) -> list[str]: +def _section_resources_gpu(results_dir: Path, slug: str = "") -> list[str]: """Second side: optional GPU campaign rows (never mixed with CPU stats).""" from benchmarks.scalability.results_schema import load_rows - res_file = results_dir / "resources_gpu.jsonl" + sfx = f"_{slug}" if slug else "" + tag = f" [{slug}]" if slug else "" + res_file = results_dir / f"resources{sfx}_gpu.jsonl" if not res_file.exists(): return [] rows_all = load_rows(res_file) ok = [r for r in rows_all if r["status"] == "ok"] out = [ - "## Resource consumption — GPU (device=cuda)", "", + "## Resource consumption — GPU (device=cuda)" + tag, "", "Fixed workloads, device GPU (row `device=cuda`). Every statistic " "below is computed from GPU rows only and never combined with the " "CPU side above. CPU% = sum over the live process tree; RSS summed " @@ -213,17 +217,19 @@ def _section_resources_gpu(results_dir: Path) -> list[str]: out += [ f"- GPU telemetry (device side cuda): {'available (device-level)' if gpu.get('available') else 'unavailable — ' + str(gpu.get('reason', ''))}.", f"- Energy (device side cuda): {'available — ' + str(energy.get('scope', '')) if energy.get('available') else 'unavailable — ' + str(energy.get('reason', ''))}.", - "- Raw per-run rows: `results/resources_gpu.jsonl`.", "", + f"- Raw per-run rows: `results/resources{sfx}_gpu.jsonl`.", "", ] out += _memory_subsections(ok) return out -def generate_report(results_dir: Path, out_path: Path) -> Path: - body = _section_resources(results_dir) - gpu_body = _section_resources_gpu(results_dir) +def generate_report(results_dir: Path, out_path: Path, + slug: str = "") -> Path: + body = _section_resources(results_dir, slug) + gpu_body = _section_resources_gpu(results_dir, slug) if not body and not gpu_body: - body = ["## Resource consumption (experiment R)", "", + body = ["## Resource consumption (experiment R)" + + (f" [{slug}]" if slug else ""), "", "_no ok rows yet._", ""] lines = [ "# Resource-consumption benchmark report", @@ -246,6 +252,9 @@ def build_parser() -> argparse.ArgumentParser: default=str(REPO_ROOT / "benchmarks/resources/results")) p.add_argument("--out", default=None, help="output markdown (default: /REPORT.md)") + p.add_argument("--env-slug", default="", + help="env slug used in result filenames " + "(default '': reads resources.jsonl, unchanged)") return p @@ -253,7 +262,7 @@ def main(argv: list[str] | None = None) -> int: args = build_parser().parse_args(argv) results_dir = Path(args.results_dir) out = Path(args.out) if args.out else results_dir / "REPORT.md" - generate_report(results_dir, out) + generate_report(results_dir, out, slug=args.env_slug) return 0 diff --git a/benchmarks/resources/run.py b/benchmarks/resources/run.py index 6b671bc..e8efcde 100644 --- a/benchmarks/resources/run.py +++ b/benchmarks/resources/run.py @@ -62,11 +62,16 @@ import psutil -from benchmarks.scalability.results_schema import append_row, make_row +from benchmarks.scalability.results_schema import ( # noqa: E402 + append_row, + env_slug, + make_row, +) REPO_ROOT = Path(__file__).resolve().parents[2] FRAMEWORKS_R = ("zeroRL", "SB3", "TorchRL") WORKLOADS = ("collect", "ppo") +DEFAULT_ENV = "LunarLander-v3" DEFAULT_INTERVAL_S = 0.2 COLLECT_NUM_ENVS = 8 PPO_EXPERIMENT = "D" @@ -646,9 +651,12 @@ def run_resource_workload( if write_samples and mon.samples: sdir = out.parent / "samples" sdir.mkdir(parents=True, exist_ok=True) - # `_cuda` suffix keeps GPU samples from overwriting CPU sample files + # `_cuda` suffix keeps GPU samples from overwriting CPU sample files; + # a non-default env gets its slug prefix so campaigns can't collide. + slug_pfx = (f"{env_slug(cfg.env_id)}_" + if cfg.env_id != DEFAULT_ENV else "") sfile = sdir / ( - f"{workload}_{framework}_rep{rep}" + f"{slug_pfx}{workload}_{framework}_rep{rep}" f"{'_cuda' if device == 'cuda' else ''}.jsonl") sfile.write_text( "\n".join(json.dumps(s) for s in mon.samples) + "\n", @@ -727,16 +735,16 @@ def run_resource_workload( return row -def _collect_cfg() -> Any: +def _collect_cfg(env_id: str = DEFAULT_ENV) -> Any: from benchmarks.scalability.collection import CollectionConfig - return CollectionConfig() + return CollectionConfig(env_id=env_id) -def _ppo_cfg() -> Any: +def _ppo_cfg(env_id: str = DEFAULT_ENV) -> Any: from benchmarks.scalability.ppo_protocols import PPOProtocolConfig - return PPOProtocolConfig() + return PPOProtocolConfig(env_id=env_id) # --------------------------------------------------------------------------- @@ -807,9 +815,13 @@ def build_parser() -> argparse.ArgumentParser: help="repetitions for the collect workload (A schedule)") p.add_argument("--reps-ppo", type=int, default=3, help="repetitions for the ppo workload (C/D schedule)") + p.add_argument("--env", default=DEFAULT_ENV, + help=f"gymnasium env id (default {DEFAULT_ENV}; a " + "non-default env defaults to resources_.jsonl)") p.add_argument("--out", default=None, help="JSONL output (default: " - "benchmarks/resources/results/resources.jsonl)") + "benchmarks/resources/results/resources.jsonl, " + "with _ appended for non-default --env)") p.add_argument("--device", choices=["cpu", "cuda"], default="cpu", help="compute device for the workload (default cpu; " "env stepping is always CPU)") @@ -822,7 +834,12 @@ def build_parser() -> argparse.ArgumentParser: def main(argv: list[str] | None = None) -> int: args = build_parser().parse_args(argv) - out = Path(args.out) if args.out else DEFAULT_OUT + if args.out: + out = Path(args.out) + elif args.env != DEFAULT_ENV: + out = DEFAULT_OUT.with_name(f"resources_{env_slug(args.env)}.jsonl") + else: + out = DEFAULT_OUT if args.validate: summary = run_validation() print(json.dumps(summary, indent=2)) @@ -843,11 +860,14 @@ def main(argv: list[str] | None = None) -> int: reps = args.reps_collect if workload == "collect" else args.reps_ppo for fw in frameworks: for rep in range(reps): - label = f"{fw:>7} {workload:>7} rep={rep} seed={1000 + rep}" + label = (f"{fw:>7} {workload:>7} env={args.env} " + f"rep={rep} seed={1000 + rep}") try: row = run_resource_workload( fw, workload, rep, interval=args.interval, out_path=out, + collection_cfg=_collect_cfg(args.env), + ppo_cfg=_ppo_cfg(args.env), write_samples=not args.no_samples, device=args.device) except Exception: diff --git a/benchmarks/scalability/collection.py b/benchmarks/scalability/collection.py index 4060c45..5476312 100644 --- a/benchmarks/scalability/collection.py +++ b/benchmarks/scalability/collection.py @@ -43,10 +43,13 @@ from benchmarks.scalability.budget import BudgetTracker, remaining_batches from benchmarks.scalability.policy_reference import ( + ACT_DIM, + OBS_DIM, TorchRLSharedTrunk, build_deterministic_zerorl_agent, build_reference_state_dict, build_sb3_shared_policy, + env_dims, load_into_sb3, ) @@ -62,7 +65,7 @@ class CollectionConfig: """Workload definition for Experiment A (shared by all frameworks).""" - env_id: str = "LunarLander-v3" + env_id: str = "HumanoidStandup-v5" warmup: int = 16_000 measure: int = 100_000 reference_seed: int = 4242 @@ -179,10 +182,12 @@ def _make_fingerprint_part() -> dict[str, Any]: def _absorb_actions(fb: dict[str, Any], actions: list[int], - tracker: BudgetTracker) -> None: + tracker: BudgetTracker, *, + is_discrete: bool = True) -> None: for a in actions: if len(fb["actions_head"]) < FINGERPRINT_HEAD: - fb["actions_head"].append(int(a)) + fb["actions_head"].append(int(a) if is_discrete + else round(float(a), 6)) total = int(sum(actions)) if tracker.target == fb.get("_warm_target"): fb["action_sum_warmup"] += total @@ -323,9 +328,12 @@ def run_zerorl_collection(cfg: CollectionConfig, num_envs: int, seed: int, factory = seeded_env_factory(cfg.env_id, seeds) env = vectorize_env(factory, num_envs=num_envs) lc.add("env", env.close) # P6: never skip env close - obs_shape, act_shape, _, _, _ = get_obs_act(env) + obs_shape, act_shape, obs_n, act_n, is_discrete = get_obs_act(env) agent = build_deterministic_zerorl_agent( - build_reference_state_dict(cfg.reference_seed) + build_reference_state_dict(cfg.reference_seed, obs_dim=obs_n, + act_dim=act_n, + is_discrete=is_discrete), + obs_dim=obs_n, act_dim=act_n, is_discrete=is_discrete, ) agent.to(device) buffer = get_actor_critic_buffer( @@ -377,7 +385,8 @@ def run_zerorl_collection(cfg: CollectionConfig, num_envs: int, seed: int, warm_n = cfg.warmup fb = { "first_obs_head": [round(float(x), 6) for x in first_obs.reshape(-1)[:4]], - "actions_head": [int(a) for a in actions[:FINGERPRINT_HEAD]], + "actions_head": [int(a) if is_discrete else round(float(a), 6) + for a in actions[:FINGERPRINT_HEAD]], "action_sum_warmup": float(np.sum(actions[:warm_n])), "action_sum_measure": float(np.sum(actions[warm_n:])), } @@ -433,8 +442,19 @@ def run_sb3_collection(cfg: CollectionConfig, num_envs: int, seed: int, vec_cls = SubprocVecEnv if subproc else DummyVecEnv vec = vec_cls(seeded_env_fns(cfg.env_id, seeds)) lc.add("vec_env", vec.close) # P6: terminates SubprocVecEnv workers - policy = build_sb3_shared_policy(vec.observation_space, vec.action_space) - load_into_sb3(policy, build_reference_state_dict(cfg.reference_seed)) + obs_space = vec.observation_space + act_space = vec.action_space + is_discrete = isinstance(act_space, gym.spaces.Discrete) + obs_n = int(np.prod(obs_space.shape)) + if is_discrete: + act_n, action_shape = int(act_space.n), () + else: + act_n = int(np.prod(act_space.shape)) + action_shape = tuple(act_space.shape) + policy = build_sb3_shared_policy(obs_space, act_space) + load_into_sb3(policy, build_reference_state_dict( + cfg.reference_seed, obs_dim=obs_n, act_dim=act_n, + is_discrete=is_discrete)) policy.to(device) # obs_to_tensor/fwd follow policy.device; buffer stays CPU # Native rollout storage: rows are env-steps; +1 slack so SB3's # pos-wrap at exact capacity can never eat a counted row. @@ -454,7 +474,8 @@ def run_sb3_collection(cfg: CollectionConfig, num_envs: int, seed: int, rollout_buf.reset() t = time.perf_counter() obs = _sb3_collect_phase(vec, policy, rollout_buf, entropy_side, obs, - warm_steps, num_envs, warm_tracker, fb) + warm_steps, num_envs, warm_tracker, fb, + action_shape=action_shape) warm_s = time.perf_counter() - t _verify_transition_count(warm_tracker.actual, cfg.warmup, "warmup") @@ -467,7 +488,8 @@ def run_sb3_collection(cfg: CollectionConfig, num_envs: int, seed: int, rollout_buf.reset() t = time.perf_counter() obs = _sb3_collect_phase(vec, policy, rollout_buf, entropy_side, obs, - meas_steps, num_envs, meas_tracker, fb) + meas_steps, num_envs, meas_tracker, fb, + action_shape=action_shape) meas_s = time.perf_counter() - t _verify_transition_count(meas_tracker.actual, cfg.measure, "measure") # Measure-phase fields come from what the buffer actually stored @@ -500,18 +522,25 @@ def run_sb3_collection(cfg: CollectionConfig, num_envs: int, seed: int, def _sb3_collect_phase(vec, policy, rollout_buf, entropy_side, obs, steps: int, num_envs: int, tracker: BudgetTracker, - fb: dict[str, Any]): + fb: dict[str, Any], + action_shape: tuple[int, ...] = ()): """One SB3 collection phase: single-pass forward producing all four Exp A fields, stored in SB3's native RolloutBuffer (+ entropy side array). - Returns the final observation (pre-boundary).""" + ``action_shape`` is the PER-ENV vec.step action shape (``()`` for + Discrete, e.g. ``(17,)`` for Box). Returns the final observation + (pre-boundary).""" prev_dones = np.zeros(num_envs, dtype=np.bool_) + is_discrete = not action_shape # () per-env shape <=> Discrete actions for _ in range(steps): with torch.no_grad(): obs_t, _ = policy.obs_to_tensor(obs) actions_t, values, log_probs, entropy = \ policy.forward_collection(obs_t) - actions = actions_t.cpu().numpy().reshape(-1) - _absorb_actions(fb, np.asarray(actions).reshape(-1).tolist(), tracker) + actions = actions_t.cpu().numpy() + actions = (actions.reshape(num_envs, *action_shape) if action_shape + else actions.reshape(num_envs)) + _absorb_actions(fb, np.asarray(actions).reshape(-1).tolist(), tracker, + is_discrete=is_discrete) next_obs, rewards, dones, _ = vec.step(actions) row = rollout_buf.pos entropy_side[row] = entropy.cpu().numpy().reshape(num_envs) @@ -550,7 +579,15 @@ def run_torchrl_collection(cfg: CollectionConfig, num_envs: int, seed: int, # GymWrapper's info-reader check consumed the canonical first reset; # re-arm so the collector's carrier reset below IS the canonical reset. rearm_first_reset(venv) - trunk = TorchRLSharedTrunk(build_reference_state_dict(cfg.reference_seed)) + d = env_dims(cfg.env_id) + sd = build_reference_state_dict(cfg.reference_seed, obs_dim=d.obs_n, + act_dim=d.act_n, + is_discrete=d.is_discrete) + # default (LunarLander) path keeps the ORIGINAL call signature + if (d.obs_n, d.act_n, d.is_discrete) == (OBS_DIM, ACT_DIM, True): + trunk = TorchRLSharedTrunk(sd) + else: + trunk = TorchRLSharedTrunk(sd, obs_dim=d.obs_n, act_dim=d.act_n) trunk.to(device) # params match policy_device => collector shares them policy = _TorchRLCollectPolicy(trunk) tdm = TensorDictModule(policy, in_keys=["observation"], @@ -580,7 +617,7 @@ def run_torchrl_collection(cfg: CollectionConfig, num_envs: int, seed: int, # td batch dims are (env, time); fingerprint order must be # time-major (step0_env0, step0_env1, ...) to match zeroRL/SB3. _absorb_actions(fb, td["action"].movedim(-1, 0).reshape(-1).tolist(), - warm_tracker) + warm_tracker, is_discrete=d.is_discrete) warm_tracker.record(n) warm_s = time.perf_counter() - t _verify_transition_count(warm_tracker.actual, cfg.warmup, "warmup") @@ -604,7 +641,7 @@ def run_torchrl_collection(cfg: CollectionConfig, num_envs: int, seed: int, for td in collector2: n = int(td.batch_size.numel()) _absorb_actions(fb, td["action"].movedim(-1, 0).reshape(-1).tolist(), - meas_tracker) + meas_tracker, is_discrete=d.is_discrete) for key in meas_fields: meas_fields[key] += n meas_tracker.record(n) @@ -645,6 +682,8 @@ class _TorchRLCollectPolicy(torch.nn.Module): def __init__(self, trunk: TorchRLSharedTrunk) -> None: super().__init__() self.trunk = trunk + # presence of the trainable std parameter <=> continuous (Box) policy + self.continuous = hasattr(trunk, "log_std") def forward(self, obs: torch.Tensor) -> tuple[torch.Tensor, ...]: # ONE shared-trunk execution per step: actor_path/critic_path would @@ -652,6 +691,17 @@ def forward(self, obs: torch.Tensor) -> tuple[torch.Tensor, ...]: features = self.trunk.trunk(obs) logits = self.trunk.actor_head(features) value = self.trunk.critic_head(features).squeeze(-1) + if self.continuous: + # Mirrors zeroRL build_distribution/eval_action exactly: + # scale = exp(clamp(log_std, -3, 1)), deterministic action = + # Normal mode (mean), log_prob/entropy summed over the action dim. + scale = torch.clamp(self.trunk.log_std, min=-3.0, + max=1.0).exp().expand_as(logits) + dist = torch.distributions.Normal(logits, scale) + actions = logits + log_prob = dist.log_prob(actions).sum(dim=-1) + entropy = dist.entropy().sum(dim=-1) + return actions, log_prob, entropy, value dist = torch.distributions.Categorical(logits=logits) actions = dist.probs.argmax(dim=-1) log_prob = dist.log_prob(actions) diff --git a/benchmarks/scalability/plots.py b/benchmarks/scalability/plots.py index b9817e0..f761fe2 100644 --- a/benchmarks/scalability/plots.py +++ b/benchmarks/scalability/plots.py @@ -1,5 +1,4 @@ """Generate benchmark figures (PNG) from the result JSONL files. - Every figure is derived from aggregated *all-repetition* statistics (mean ± std); nothing is hand-picked. Experiments with no result file (or a missing baseline point, e.g. while runs are still in flight) are skipped with a @@ -238,31 +237,35 @@ def plot_ppo_experiment(rows, experiment: str, out_dir: Path, return paths -def generate_all(results_dir: Path, out_dir: Path | None = None) -> list[Path]: +def generate_all(results_dir: Path, out_dir: Path | None = None, + slug: str = "") -> list[Path]: out_dir = out_dir or results_dir / "figures" + sfx = f"_{slug}" if slug else "" paths: list[Path] = [] for exp, plotter in ( ("A", plot_experiment_a), ("B", plot_experiment_b), ): - file = results_dir / f"exp{exp}.jsonl" + file = results_dir / f"exp{exp}{sfx}.jsonl" if not file.exists(): warnings.warn(f"{file} missing — skipping exp {exp}") else: - paths.extend(plotter(load_flat(file), out_dir)) - gpu_file = results_dir / f"exp{exp}_gpu.jsonl" # silent when absent + paths.extend(plotter(load_flat(file), out_dir, suffix=sfx)) + gpu_file = results_dir / f"exp{exp}{sfx}_gpu.jsonl" # silent when absent if gpu_file.exists(): - paths.extend(plotter(load_flat(gpu_file), out_dir, suffix="_gpu")) + paths.extend(plotter(load_flat(gpu_file), out_dir, + suffix=f"{sfx}_gpu")) for exp in ("C", "D"): - file = results_dir / f"exp{exp}.jsonl" + file = results_dir / f"exp{exp}{sfx}.jsonl" if not file.exists(): warnings.warn(f"{file} missing — skipping exp {exp}") else: - paths.extend(plot_ppo_experiment(load_flat(file), exp, out_dir)) - gpu_file = results_dir / f"exp{exp}_gpu.jsonl" # silent when absent + paths.extend(plot_ppo_experiment(load_flat(file), exp, out_dir, + suffix=sfx)) + gpu_file = results_dir / f"exp{exp}{sfx}_gpu.jsonl" # silent when absent if gpu_file.exists(): paths.extend(plot_ppo_experiment(load_flat(gpu_file), exp, out_dir, - suffix="_gpu")) + suffix=f"{sfx}_gpu")) return paths @@ -272,10 +275,13 @@ def main(argv: list[str] | None = None) -> int: default=str(REPO_ROOT / "benchmarks/scalability/results")) p.add_argument("--out", default=None, help="figure output dir (default: /figures)") + p.add_argument("--env-slug", default="", + help="env slug used in result/figure filenames " + "(default '': reads exp{X}.jsonl, unchanged)") args = p.parse_args(argv) results_dir = Path(args.results_dir) out_dir = Path(args.out) if args.out else None - paths = generate_all(results_dir, out_dir) + paths = generate_all(results_dir, out_dir, slug=args.env_slug) if not paths: print("no figures generated (no result files found?)") return 1 diff --git a/benchmarks/scalability/policy_reference.py b/benchmarks/scalability/policy_reference.py index d8122ea..73c8798 100644 --- a/benchmarks/scalability/policy_reference.py +++ b/benchmarks/scalability/policy_reference.py @@ -19,8 +19,10 @@ from __future__ import annotations -from typing import Any +from functools import lru_cache +from typing import Any, NamedTuple +import gymnasium as gym import numpy as np import torch import torch.nn as nn @@ -46,15 +48,65 @@ ) -def build_reference_state_dict(seed: int = REFERENCE_SEED) -> dict[str, torch.Tensor]: +class EnvDims(NamedTuple): + """Space dimensions of one env, mirroring ``get_obs_act`` semantics. + + ``obs_n`` is the flat observation size fed to the MLP (the space itself + for 3-D image observations, as ``get_obs_act`` returns); ``act_n`` is + the Discrete ``n`` or the product of the Box shape; ``act_shape`` is + ``()`` for Discrete. + """ + + obs_shape: tuple[int, ...] + act_shape: tuple[int, ...] + obs_n: Any + act_n: int + is_discrete: bool + + +@lru_cache(maxsize=None) +def env_dims(env_id: str) -> EnvDims: + """Probe an env's spaces (``gym.make`` -> read -> close); lru-cached. + + Mirrors ``zerorl.functions.get_obs_act`` exactly: + discrete -> ``act_shape=()``, ``act_n=space.n``; Box -> ``act_shape= + space.shape``, ``act_n=int(np.prod(shape))``; ``obs_n = shape[-1]`` + (or the space itself for 3-D image observations). + """ + env = gym.make(env_id) + try: + obs_space = env.observation_space + act_space = env.action_space + finally: + env.close() + obs_shape = tuple(obs_space.shape) + act_shape = tuple(act_space.shape) + is_discrete = isinstance(act_space, gym.spaces.Discrete) + if is_discrete: + act_n = int(act_space.n) # type: ignore[attr-defined] + else: + act_n = int(np.prod(act_space.shape)) # type: ignore[arg-type] + obs_n: Any = obs_space if len(obs_shape) == 3 else obs_shape[-1] + return EnvDims(obs_shape, act_shape, obs_n, act_n, is_discrete) + + +def build_reference_state_dict(seed: int = REFERENCE_SEED, *, + obs_dim: int = OBS_DIM, + act_dim: int = ACT_DIM, + is_discrete: bool = True, + ) -> dict[str, torch.Tensor]: """Deterministic reference weights via zeroRL's own agent init. Uses ``ActorCriticAgent`` so the init scheme (ortho gains √2 trunk / 0.01 actor / 1.0 critic, zero biases) and key layout are exactly zeroRL's defaults — the other frameworks map these tensors onto their modules. + The default (LunarLander-shaped, discrete) call is byte-identical to the + original signature. Continuous agents additionally carry a ``log_std`` + key (extra keys are allowed; REFERENCE_KEYS are only the required ones). """ torch.manual_seed(seed) - agent = ActorCriticAgent(OBS_DIM, ACT_DIM, is_discrete=True, hidden_dim=HIDDEN) + agent = ActorCriticAgent(obs_dim, act_dim, is_discrete=is_discrete, + hidden_dim=HIDDEN) sd = {k: v.detach().clone() for k, v in agent.state_dict().items()} missing = [k for k in REFERENCE_KEYS if k not in sd] if missing: @@ -94,10 +146,13 @@ def get_action(self, state: torch.Tensor, action: torch.Tensor | None = None, **kwargs: Any) -> dict[str, torch.Tensor]: logits, value = self.forward(state, **kwargs) dist = self.build_distribution(logits) - greedy = logits.argmax(dim=-1) - log_prob, entropy = eval_action(dist, greedy) + if self.is_discrete: + taken = logits.argmax(dim=-1) + else: + taken = logits # Normal mode = mean (deterministic) + log_prob, entropy = eval_action(dist, taken) return { - "action": greedy, + "action": taken, "log_prob": log_prob, "entropy": entropy, "value": value.squeeze(-1), @@ -106,10 +161,19 @@ def get_action(self, state: torch.Tensor, action: torch.Tensor | None = None, def build_deterministic_zerorl_agent( sd: dict[str, torch.Tensor] | None = None, + *, + obs_dim: int = OBS_DIM, + act_dim: int = ACT_DIM, + is_discrete: bool = True, ) -> DeterministicActorCriticAgent: - agent = DeterministicActorCriticAgent(OBS_DIM, ACT_DIM, is_discrete=True, + agent = DeterministicActorCriticAgent(obs_dim, act_dim, + is_discrete=is_discrete, hidden_dim=HIDDEN) - load_into_zerorl(agent, sd if sd is not None else build_reference_state_dict()) + load_into_zerorl( + agent, + sd if sd is not None else build_reference_state_dict( + obs_dim=obs_dim, act_dim=act_dim, is_discrete=is_discrete), + ) return agent @@ -248,6 +312,16 @@ def load_into_sb3(policy: nn.Module, sd: dict[str, torch.Tensor]) -> None: "critic.weight": policy.value_net.weight, # type: ignore[attr-defined] "critic.bias": policy.value_net.bias, } + if "log_std" in sd: + # Box policies own a separate Gaussian log_std Parameter + # (sb3 policies.py: `self.action_net, self.log_std = ...proba_distribution_net`) + if not hasattr(policy, "log_std"): + raise AssertionError( + "reference sd carries log_std (continuous) but the SB3 " + "policy has no log_std parameter (not a Box policy?)") + with torch.no_grad(): + policy.log_std.copy_(sd["log_std"]) # type: ignore[attr-defined] + read_back["log_std"] = policy.log_std # type: ignore[attr-defined] _assert_equal(read_back, sd, "SB3 policy") @@ -255,6 +329,24 @@ def load_into_sb3(policy: nn.Module, sd: dict[str, torch.Tensor]) -> None: # TorchRL adapter # --------------------------------------------------------------------------- +class _TrunkSequential(nn.Sequential): + """Shared trunk that EXECUTES in float32. + + zeroRL casts observations to float32 at reset (``trainer.state = + torch.as_tensor(..., dtype=torch.float32)``) and SB3 casts in its + obs-to-tensor path; gymnasium MuJoCo envs (e.g. HumanoidStandup-v4) + emit float64 observations, so the TorchRL path — which feeds raw env + tensors into the policy — casts at the trunk boundary to keep policy + execution dtype-identical across frameworks. No-op for float32 inputs + (e.g. LunarLander), so the discrete path is numerically unchanged. + """ + + def forward(self, input: torch.Tensor) -> torch.Tensor: # noqa: A002 + if input.dtype != torch.float32: + input = input.to(torch.float32) # noqa: A001 + return super().forward(input) + + class TorchRLSharedTrunk(nn.Module): """Shared 64-64 Tanh trunk + actor/critic heads (zeroRL topology). @@ -263,17 +355,40 @@ class TorchRLSharedTrunk(nn.Module): ``parameters()`` — matching zeroRL's single-shared-trunk agent exactly. """ - def __init__(self, sd: dict[str, torch.Tensor] | None = None) -> None: + def __init__(self, sd: dict[str, torch.Tensor] | None = None, *, + obs_dim: int = OBS_DIM, act_dim: int = ACT_DIM) -> None: super().__init__() - self.trunk = nn.Sequential( - nn.Linear(OBS_DIM, HIDDEN), + self.trunk = _TrunkSequential( + nn.Linear(obs_dim, HIDDEN), nn.Tanh(), nn.Linear(HIDDEN, HIDDEN), nn.Tanh(), ) - self.actor_head = nn.Linear(HIDDEN, ACT_DIM) + self.actor_head = nn.Linear(HIDDEN, act_dim) self.critic_head = nn.Linear(HIDDEN, 1) - sd = sd if sd is not None else build_reference_state_dict() + sd = sd if sd is not None else build_reference_state_dict( + obs_dim=obs_dim, act_dim=act_dim) + # clear failure (not a raw copy_ RuntimeError) on a dims mismatch + if (tuple(sd["actor.weight"].shape) != (act_dim, HIDDEN) + or tuple(sd["extract_layer.0.weight"].shape) != (HIDDEN, obs_dim)): + raise ValueError( + f"TorchRLSharedTrunk dims mismatch: obs_dim={obs_dim}, " + f"act_dim={act_dim} require actor.weight " + f"{(act_dim, HIDDEN)} and extract_layer.0.weight " + f"{(HIDDEN, obs_dim)}, got " + f"{tuple(sd['actor.weight'].shape)} and " + f"{tuple(sd['extract_layer.0.weight'].shape)}" + ) + if "log_std" in sd: + # continuous policy std: TRAINABLE (zeroRL optimizes it; picked + # up by parameters()/_dedup_params) and shape-checked vs dims + if tuple(sd["log_std"].shape) != (act_dim,): + raise ValueError( + f"TorchRLSharedTrunk dims mismatch: act_dim={act_dim} " + f"requires log_std shape {(act_dim,)}, got " + f"{tuple(sd['log_std'].shape)}" + ) + self.log_std = nn.Parameter(sd["log_std"].clone()) with torch.no_grad(): self.trunk[0].weight.copy_(sd["extract_layer.0.weight"]) self.trunk[0].bias.copy_(sd["extract_layer.0.bias"]) diff --git a/benchmarks/scalability/ppo_protocols.py b/benchmarks/scalability/ppo_protocols.py index b0d05ec..c268550 100644 --- a/benchmarks/scalability/ppo_protocols.py +++ b/benchmarks/scalability/ppo_protocols.py @@ -57,6 +57,7 @@ OBS_DIM, TorchRLSharedTrunk, build_reference_state_dict, + env_dims, load_into_sb3, load_into_zerorl, shared_trunk_policy_cls, @@ -70,7 +71,7 @@ class PPOProtocolConfig: """Workload + shared hyperparameters for Experiments C/D.""" - env_id: str = "LunarLander-v3" + env_id: str = "HumanoidStandup-v5" total: int = DEFAULT_PPO_TOTAL lr: float = 3e-4 gamma: float = 0.99 @@ -198,9 +199,12 @@ def run_zerorl_ppo(cfg: PPOProtocolConfig, experiment: str, num_envs: int, with lifecycle() as lc: t0 = time.perf_counter() - agent = ActorCriticAgent(OBS_DIM, ACT_DIM, is_discrete=True, - hidden_dim=HIDDEN) - load_into_zerorl(agent, build_reference_state_dict(cfg.reference_seed)) + d = env_dims(cfg.env_id) + agent = ActorCriticAgent(d.obs_n, d.act_n, + is_discrete=d.is_discrete, hidden_dim=HIDDEN) + load_into_zerorl(agent, build_reference_state_dict( + cfg.reference_seed, obs_dim=d.obs_n, act_dim=d.act_n, + is_discrete=d.is_discrete)) agent.to(device) tconfig = TrainConfig( model_name=f"bench_ppo_{experiment}", @@ -289,7 +293,10 @@ def run_sb3_ppo(cfg: PPOProtocolConfig, experiment: str, num_envs: int, verbose=0, device=device, # SB3 keeps its native rollout buffer on CPU ) - load_into_sb3(model.policy, build_reference_state_dict(cfg.reference_seed)) + d = env_dims(cfg.env_id) + load_into_sb3(model.policy, build_reference_state_dict( + cfg.reference_seed, obs_dim=d.obs_n, act_dim=d.act_n, + is_discrete=d.is_discrete)) init_s = time.perf_counter() - t0 lr_start, opt_ev = _lr_and_optimizer(model.policy.optimizer) @@ -357,9 +364,41 @@ def _dedup_params(*modules: torch.nn.Module) -> list[torch.nn.Parameter]: return out +class SummedNormal(torch.distributions.Normal): + """Normal with log_prob/entropy summed over the action dim. + + Exactly zeroRL's ``eval_action`` reduction (``sum(dim=-1)`` for 2-D + outputs). torchrl's plain ``Normal`` leaves per-action-dim (B, A) + values for Tensor actions (only CompositeDistribution aggregates), + which would mis-shape ClipPPOLoss vs the working (B,) Categorical + path — this subclass keeps both paths shape-identical. + """ + + def log_prob(self, action: torch.Tensor) -> torch.Tensor: + return super().log_prob(action).sum(dim=-1) + + def entropy(self) -> torch.Tensor: + return super().entropy().sum(dim=-1) + + +class _MeanScaleHead(torch.nn.Module): + """Continuous PPO actor params: mean from the shared actor head, scale + from the trunk's TRAINABLE log_std with zeroRL's clamp semantics.""" + + def __init__(self, trunk: TorchRLSharedTrunk) -> None: + super().__init__() + self.trunk = trunk + + def forward(self, features: torch.Tensor) -> tuple[torch.Tensor, torch.Tensor]: + mean = self.trunk.actor_head(features) + scale = self.trunk.log_std.clamp(-3.0, 1.0).exp().expand_as(mean) + return mean, scale + + def build_torchrl_ppo_wiring(trunk: TorchRLSharedTrunk, cfg: PPOProtocolConfig, - updates: int) -> dict[str, Any]: + updates: int, + is_discrete: bool = True) -> dict[str, Any]: """Single-shared-trunk PPO wiring for TorchRL (shared-trunk fix). One trunk execution per actor-critic pass: @@ -373,6 +412,11 @@ def build_torchrl_ppo_wiring(trunk: TorchRLSharedTrunk, - GAE still evaluates the critic from ``td["features"]`` / ``td["next","features"]`` (targets, no grad; next-features injected per batch in the runner). + + Action space: ``is_discrete=True`` wires Categorical over logits; + False wires :class:`SummedNormal` over (mean, scale) emitted by + :class:`_MeanScaleHead` (dict ``in_keys`` map Normal's ``loc``/``scale`` + kwargs onto the td keys, per tensordict's ProbabilisticActor API). """ from tensordict.nn import TensorDictModule, TensorDictSequential, composite_lp_aggregate from torchrl.modules import ProbabilisticActor, ValueOperator @@ -383,24 +427,44 @@ def build_torchrl_ppo_wiring(trunk: TorchRLSharedTrunk, else "action_log_prob") feat_td = TensorDictModule(trunk.trunk, in_keys=["observation"], out_keys=["features"]) - collect_logits = TensorDictModule(trunk.actor_head, - in_keys=["features"], - out_keys=["logits"]) - loss_logits = TensorDictModule(trunk.actor_head, - in_keys=["features"], - out_keys=["logits"]) - collector_policy = ProbabilisticActor( - module=TensorDictSequential(feat_td, collect_logits), - in_keys=["logits"], out_keys=["action"], - distribution_class=torch.distributions.Categorical, - return_log_prob=True, log_prob_key=lp_key, - ) - loss_actor = ProbabilisticActor( - module=loss_logits, - in_keys=["logits"], out_keys=["action"], - distribution_class=torch.distributions.Categorical, - return_log_prob=True, log_prob_key=lp_key, - ) + if is_discrete: + collect_logits = TensorDictModule(trunk.actor_head, + in_keys=["features"], + out_keys=["logits"]) + loss_logits = TensorDictModule(trunk.actor_head, + in_keys=["features"], + out_keys=["logits"]) + collector_policy = ProbabilisticActor( + module=TensorDictSequential(feat_td, collect_logits), + in_keys=["logits"], out_keys=["action"], + distribution_class=torch.distributions.Categorical, + return_log_prob=True, log_prob_key=lp_key, + ) + loss_actor = ProbabilisticActor( + module=loss_logits, + in_keys=["logits"], out_keys=["action"], + distribution_class=torch.distributions.Categorical, + return_log_prob=True, log_prob_key=lp_key, + ) + else: + collect_params = TensorDictModule(_MeanScaleHead(trunk), + in_keys=["features"], + out_keys=["mean", "scale"]) + loss_params = TensorDictModule(_MeanScaleHead(trunk), + in_keys=["features"], + out_keys=["mean", "scale"]) + collector_policy = ProbabilisticActor( + module=TensorDictSequential(feat_td, collect_params), + in_keys={"loc": "mean", "scale": "scale"}, out_keys=["action"], + distribution_class=SummedNormal, + return_log_prob=True, log_prob_key=lp_key, + ) + loss_actor = ProbabilisticActor( + module=loss_params, + in_keys={"loc": "mean", "scale": "scale"}, out_keys=["action"], + distribution_class=SummedNormal, + return_log_prob=True, log_prob_key=lp_key, + ) critic = ValueOperator(module=trunk.critic_head, in_keys=["features"], out_keys=["state_value"]) @@ -470,10 +534,25 @@ def run_torchrl_ppo(cfg: PPOProtocolConfig, experiment: str, num_envs: int, from benchmarks.scalability.collection import rearm_first_reset rearm_first_reset(venv) - trunk = TorchRLSharedTrunk(build_reference_state_dict(cfg.reference_seed)) + d = env_dims(cfg.env_id) + sd = build_reference_state_dict(cfg.reference_seed, obs_dim=d.obs_n, + act_dim=d.act_n, + is_discrete=d.is_discrete) + # default (LunarLander) path keeps the ORIGINAL call signatures so + # existing test doubles/spies match unchanged + default_dims = (d.obs_n, d.act_n, d.is_discrete) == ( + OBS_DIM, ACT_DIM, True) + if default_dims: + trunk = TorchRLSharedTrunk(sd) + else: + trunk = TorchRLSharedTrunk(sd, obs_dim=d.obs_n, act_dim=d.act_n) trunk.to(device) # before wiring + Collector: params match # policy_device so the collector shares these exact params (no copy). - wiring = build_torchrl_ppo_wiring(trunk, cfg, updates) + if default_dims: + wiring = build_torchrl_ppo_wiring(trunk, cfg, updates) + else: + wiring = build_torchrl_ppo_wiring(trunk, cfg, updates, + is_discrete=d.is_discrete) lr_start, opt_ev = _lr_and_optimizer(wiring["optimizer"]) # P5: collector construction (env carrier reset + spec wiring) is # setup work — timed under init, never inside the measure timer. diff --git a/benchmarks/scalability/report.py b/benchmarks/scalability/report.py index 9b22097..3d76823 100644 --- a/benchmarks/scalability/report.py +++ b/benchmarks/scalability/report.py @@ -1,5 +1,4 @@ """Generate REPORT.md from the benchmark result JSONL files. - All numbers are computed from the raw rows (aggregated over every repetition, never best-of); findings sentences are derived programmatically so the report cannot claim anything the data does not show. Missing or @@ -47,13 +46,14 @@ def _device(row: dict) -> str: return dev or "cpu" -def _load_sides(results_dir: Path, exp: str) -> dict[str, list[dict]]: - """STRICT per-experiment file list (never glob): ``exp{X}.jsonl`` first, - then the optional ``exp{X}_gpu.jsonl``; missing files are silently - skipped. Rows are split by their own device value — sides other than - ``cpu``/``cuda`` are keyed by the value itself.""" +def _load_sides(results_dir: Path, exp: str, + sfx: str = "") -> dict[str, list[dict]]: + """STRICT per-experiment file list (never glob): ``exp{X}{sfx}.jsonl`` + first, then the optional ``exp{X}{sfx}_gpu.jsonl``; missing files are + silently skipped. Rows are split by their own device value — sides other + than ``cpu``/``cuda`` are keyed by the value itself.""" sides: dict[str, list[dict]] = {} - for name in (f"exp{exp}.jsonl", f"exp{exp}_gpu.jsonl"): + for name in (f"exp{exp}{sfx}.jsonl", f"exp{exp}{sfx}_gpu.jsonl"): file = results_dir / name if not file.exists(): continue @@ -374,16 +374,18 @@ def _hyperparams(rows: list[dict]) -> list[str]: _table(["key", "value"], [[k, str(v)] for k, v in cfg.items()]), ""] -def generate_report(results_dir: Path, out_path: Path) -> Path: +def generate_report(results_dir: Path, out_path: Path, + slug: str = "") -> Path: WARNINGS.clear() + sfx = f"_{slug}" if slug else "" sides_by_exp: dict[str, dict[str, list[dict]]] = {} rows: list[dict] = [] # CPU side (drives findings etc.) other_rows: dict[str, list[dict]] = {} # non-cpu sides, keyed by device for exp in ("A", "B", "C", "D"): - file = results_dir / f"exp{exp}.jsonl" + file = results_dir / f"exp{exp}{sfx}.jsonl" if not file.exists(): _warn(f"{file.name} missing — section skipped") - sides = _load_sides(results_dir, exp) + sides = _load_sides(results_dir, exp, sfx) sides_by_exp[exp] = sides rows.extend(sides.get("cpu", [])) for dev, dev_rows in sides.items(): @@ -415,35 +417,40 @@ def generate_report(results_dir: Path, out_path: Path) -> Path: "D": ("## Experiment D — PPO training protocol " "(fixed global rollout of 2048)"), } + tag = f" [{slug}]" if slug else "" for exp in ("A", "B", "C", "D"): sides = sides_by_exp[exp] gpu_sides = {d: rs for d, rs in sides.items() if d != "cpu"} # Existing CPU section first (byte-identical for CPU-only dirs). - if (results_dir / f"exp{exp}.jsonl").exists() and ( + if (results_dir / f"exp{exp}{sfx}.jsonl").exists() and ( sides.get("cpu") or not gpu_sides): if exp == "A": - table, curves["A"] = _exp_a_table(sides.get("cpu", [])) + table, curves["A"] = _exp_a_table(sides.get("cpu", []), + suffix=sfx) elif exp == "B": - table, curves["B"] = _exp_b_table(sides.get("cpu", [])) + table, curves["B"] = _exp_b_table(sides.get("cpu", []), + suffix=sfx) else: - table, curves[exp] = _exp_ppo_table(sides.get("cpu", []), exp) - lines += [titles[exp], "", table, ""] + table, curves[exp] = _exp_ppo_table(sides.get("cpu", []), exp, + suffix=sfx) + lines += [titles[exp] + tag, "", table, ""] # Sibling section(s) for every non-cpu side — same renderer, own # rows, `_gpu` figure infix; never mixed with the CPU statistics. for dev, dev_rows in gpu_sides.items(): label = "GPU" if dev == "cuda" else dev lines += [ - f"## Experiment {exp} — {label} (device={dev})", "", + f"## Experiment {exp} — {label} (device={dev})" + tag, "", f"Rows recorded with device={dev}; every statistic in this " "section is computed on this side only and never combined " "with CPU rows.", "", ] if exp == "A": - table, _ = _exp_a_table(dev_rows, suffix="_gpu") + table, _ = _exp_a_table(dev_rows, suffix=f"{sfx}_gpu") elif exp == "B": - table, _ = _exp_b_table(dev_rows, suffix="_gpu") + table, _ = _exp_b_table(dev_rows, suffix=f"{sfx}_gpu") else: - table, _ = _exp_ppo_table(dev_rows, exp, suffix="_gpu") + table, _ = _exp_ppo_table(dev_rows, exp, + suffix=f"{sfx}_gpu") lines += [table, ""] lines += _findings(curves, rows) @@ -465,10 +472,13 @@ def main(argv: list[str] | None = None) -> int: default=str(REPO_ROOT / "benchmarks/scalability/results")) p.add_argument("--out", default=None, help="output markdown (default: /REPORT.md)") + p.add_argument("--env-slug", default="", + help="env slug used in result/figure filenames " + "(default '': reads exp{X}.jsonl, unchanged)") args = p.parse_args(argv) results_dir = Path(args.results_dir) out = Path(args.out) if args.out else results_dir / "REPORT.md" - generate_report(results_dir, out) + generate_report(results_dir, out, slug=args.env_slug) return 0 diff --git a/benchmarks/scalability/results/REPORT.md b/benchmarks/scalability/results/REPORT.md index fffbcce..19418f8 100644 --- a/benchmarks/scalability/results/REPORT.md +++ b/benchmarks/scalability/results/REPORT.md @@ -1,6 +1,6 @@ # Scalability benchmark report -Generated: 2026-11-03T23:13:55.289212+00:00 · rows: 517 · figures in `figures/` +Generated: 2026-11-07T19:17:21.500812+00:00 · rows: 425 · figures in `figures/` ## Environment @@ -18,7 +18,7 @@ Generated: 2026-11-03T23:13:55.289212+00:00 · rows: 517 · figures in `figures/ | sb3 | 2.9.0 | | torchrl | 0.14.1 | | tensordict | 0.14.3 | -| git_sha | 8eb6392d7b63993aa0108ea1a6d24946bc28df8f | +| git_sha | 7813ea3a3711cfbcf6592ef8af5bb05b04fb4dc4 | ## Methodology (summary) @@ -29,76 +29,72 @@ Generated: 2026-11-03T23:13:55.289212+00:00 · rows: 517 · figures in `figures/ - Phase timing: `init` = env + policy construction + weight load (+ TorchRL collector/worker spawn); `warmup` (A/B only, excluded); `boundary` (A/B only) = warmup→measure phase reset / collector-worker restart, timed as `boundary_seconds`; `measure` = the official training/collection steady state. Full framework-specific details in `INSPECTION.md`. - Device: CPU (`CUDA_VISIBLE_DEVICES=""`); protocol C/D hyperparameters in `ppo_protocols.py` (lr 3e-4 linear-decay, gamma .99, gae_lambda .95, clip .2, entropy .01, value .5, epochs 10, batch 64, grad-clip .5). -## Experiment A — environment scaling (collection) +## Experiment A — environment scaling (collection) [humanoidstandupv5] | framework | N | throughput (trans/s) | mean time (s) | speedup (paired) | env-scaling efficiency (paired) | pairs (n) | | --- | --- | --- | --- | --- | --- | --- | -| zeroRL | 1 | 1,657.0 ± 15.6 | 60.35 | 1.00 | 1.00 | 5 | -| zeroRL | 2 | 2,415.7 ± 38.8 | 41.41 | 1.46 ± 0.02 | 0.73 ± 0.01 | 5 | -| zeroRL | 4 | 4,582.0 ± 47.1 | 21.83 | 2.75 ± 0.05 | 0.69 ± 0.01 | 5 | -| zeroRL | 8 | 4,755.4 ± 40.3 | 21.03 | 2.86 ± 0.03 | 0.36 ± 0.00 | 5 | -| zeroRL | 16 | 8,175.7 ± 127.6 | 12.23 | 4.93 ± 0.05 | 0.31 ± 0.00 | 5 | -| SB3 | 1 | 1,593.6 ± 10.5 | 62.75 | 1.00 | 1.00 | 5 | -| SB3 | 2 | 2,448.9 ± 18.0 | 40.84 | 1.54 ± 0.01 | 0.77 ± 0.01 | 5 | -| SB3 | 4 | 4,577.4 ± 62.5 | 21.85 | 2.85 ± 0.05 | 0.71 ± 0.01 | 5 | -| SB3 | 8 | 4,712.2 ± 29.3 | 21.22 | 2.96 ± 0.03 | 0.37 ± 0.00 | 5 | -| SB3 | 16 | 7,991.2 ± 124.6 | 12.52 | 5.02 ± 0.13 | 0.31 ± 0.01 | 5 | -| TorchRL | 1 | 826.2 ± 23.6 | 121.13 | 1.00 | 1.00 | 5 | -| TorchRL | 2 | 1,307.2 ± 73.6 | 76.72 | 1.53 ± 0.06 | 0.76 ± 0.03 | 5 | -| TorchRL | 4 | 2,382.4 ± 125.0 | 42.08 | 2.79 ± 0.10 | 0.70 ± 0.02 | 5 | -| TorchRL | 8 | 3,375.6 ± 98.6 | 29.65 | 4.05 ± 0.11 | 0.51 ± 0.01 | 5 | -| TorchRL | 16 | 5,785.2 ± 193.1 | 17.30 | 6.94 ± 0.36 | 0.43 ± 0.02 | 5 | +| zeroRL | 1 | 605.0 ± 11.0 | 165.33 | 1.00 | 1.00 | 5 | +| zeroRL | 2 | 806.7 ± 31.4 | 124.14 | 1.36 ± 0.01 | 0.68 ± 0.01 | 5 | +| zeroRL | 4 | 1,074.0 ± 32.2 | 93.19 | 1.79 ± 0.01 | 0.45 ± 0.00 | 5 | +| zeroRL | 8 | 1,136.9 ± 38.2 | 88.06 | 1.91 ± 0.02 | 0.24 ± 0.00 | 5 | +| zeroRL | 16 | 1,329.3 ± 33.9 | 75.28 | 2.21 ± 0.01 | 0.14 ± 0.00 | 5 | +| SB3 | 1 | 618.6 ± 10.1 | 161.71 | 1.00 | 1.00 | 5 | +| SB3 | 2 | 791.7 ± 34.1 | 126.52 | 1.33 ± 0.01 | 0.66 ± 0.00 | 5 | +| SB3 | 4 | 1,059.2 ± 29.9 | 94.48 | 1.74 ± 0.01 | 0.44 ± 0.00 | 5 | +| SB3 | 8 | 1,116.7 ± 41.6 | 89.67 | 1.86 ± 0.01 | 0.23 ± 0.00 | 5 | +| SB3 | 16 | 1,304.4 ± 33.7 | 76.71 | 2.14 ± 0.01 | 0.13 ± 0.00 | 5 | +| TorchRL | 1 | 369.2 ± 14.1 | 271.25 | 1.00 | 1.00 | 5 | +| TorchRL | 2 | 590.6 ± 19.1 | 169.50 | 1.59 ± 0.01 | 0.79 ± 0.01 | 5 | +| TorchRL | 4 | 864.7 ± 22.6 | 115.72 | 2.31 ± 0.02 | 0.58 ± 0.00 | 5 | +| TorchRL | 8 | 995.2 ± 38.1 | 100.61 | 2.66 ± 0.02 | 0.33 ± 0.00 | 5 | +| TorchRL | 16 | 1,221.5 ± 32.6 | 81.92 | 3.23 ± 0.02 | 0.20 ± 0.00 | 5 | -![throughput](figures/expA_throughput.png) +![throughput](figures/expA_humanoidstandupv5_throughput.png) -![speedup](figures/expA_speedup.png) +![speedup](figures/expA_humanoidstandupv5_speedup.png) -![efficiency](figures/expA_efficiency.png) +![efficiency](figures/expA_humanoidstandupv5_efficiency.png) -## Experiment A — GPU (device=cuda) - -Rows recorded with device=cuda; every statistic in this section is computed on this side only and never combined with CPU rows. - -| framework | N | throughput (trans/s) | mean time (s) | speedup (paired) | env-scaling efficiency (paired) | pairs (n) | -| --- | --- | --- | --- | --- | --- | --- | -| zeroRL | 1 | 1,084.8 ± 30.7 | 92.25 | 1.00 | 1.00 | 5 | -| zeroRL | 2 | 2,031.1 ± 37.8 | 49.25 | 1.85 ± 0.05 | 0.93 ± 0.02 | 5 | -| zeroRL | 4 | 3,752.1 ± 114.4 | 26.68 | 3.42 ± 0.07 | 0.86 ± 0.02 | 5 | -| zeroRL | 8 | 6,132.1 ± 209.2 | 16.33 | 5.62 ± 0.24 | 0.70 ± 0.03 | 5 | -| zeroRL | 16 | 10,055.8 ± 364.0 | 9.96 | 9.17 ± 0.29 | 0.57 ± 0.02 | 5 | -| SB3 | 1 | 1,301.8 ± 31.2 | 76.86 | 1.00 | 1.00 | 5 | -| SB3 | 2 | 2,458.5 ± 49.7 | 40.69 | 1.87 ± 0.04 | 0.93 ± 0.02 | 5 | -| SB3 | 4 | 4,491.4 ± 116.8 | 22.28 | 3.51 ± 0.08 | 0.88 ± 0.02 | 5 | -| SB3 | 8 | 7,658.3 ± 298.3 | 13.08 | 5.93 ± 0.20 | 0.74 ± 0.02 | 5 | -| SB3 | 16 | 12,179.2 ± 339.5 | 8.22 | 9.24 ± 0.39 | 0.58 ± 0.02 | 5 | -| TorchRL | 1 | 643.8 ± 15.5 | 155.42 | 1.00 | 1.00 | 5 | -| TorchRL | 2 | 1,245.6 ± 7.2 | 80.28 | 1.94 ± 0.05 | 0.97 ± 0.03 | 5 | -| TorchRL | 4 | 2,270.8 ± 56.6 | 44.06 | 3.55 ± 0.08 | 0.89 ± 0.02 | 5 | -| TorchRL | 8 | 4,015.5 ± 190.4 | 24.96 | 6.07 ± 0.44 | 0.76 ± 0.06 | 5 | -| TorchRL | 16 | 6,809.0 ± 204.9 | 14.70 | 10.39 ± 0.44 | 0.65 ± 0.03 | 5 | - -![throughput](figures/expA_gpu_throughput.png) - -![speedup](figures/expA_gpu_speedup.png) - -![efficiency](figures/expA_gpu_efficiency.png) - - -## Experiment B — worker scaling (collection, N=8) +## Experiment B — worker scaling (collection, N=8) [humanoidstandupv5] | framework | W | throughput (trans/s) | mean time (s) | speedup (paired) | worker parallel efficiency (paired) | pairs (n) | | --- | --- | --- | --- | --- | --- | --- | -| zeroRL | 1 | 4,474.1 ± 20.8 | 22.35 | 1.00 | 1.00 | 5 | -| SB3 | 1 | 4,472.1 ± 38.7 | 22.36 | 1.00 | 1.00 | 5 | -| SB3 | 8 | 3,966.0 ± 59.4 | 25.22 | 0.89 ± 0.02 | 0.11 ± 0.00 | 5 | -| TorchRL | 1 | 3,308.5 ± 8.5 | 30.23 | 1.00 | 1.00 | 5 | -| TorchRL | 2 | 7,332.2 ± 273.1 | 13.65 | 2.22 ± 0.08 | 1.11 ± 0.04 | 5 | -| TorchRL | 4 | 7,842.2 ± 69.0 | 12.75 | 2.37 ± 0.02 | 0.59 ± 0.01 | 5 | -| TorchRL | 8 | 8,452.6 ± 104.3 | 11.83 | 2.55 ± 0.04 | 0.32 ± 0.00 | 5 | +| zeroRL | 1 | 1,159.6 ± 6.7 | 86.24 | 1.00 | 1.00 | 5 | +| SB3 | 1 | 1,167.6 ± 3.3 | 85.65 | 1.00 | 1.00 | 5 | +| SB3 | 8 | 2,259.8 ± 57.7 | 44.28 | 1.97 ± 0.04 | 0.25 ± 0.01 | 5 | +| TorchRL | 1 | 1,030.9 ± 3.2 | 97.01 | 1.00 | 1.00 | 5 | +| TorchRL | 2 | 2,101.8 ± 20.2 | 47.58 | 2.04 ± 0.02 | 1.02 ± 0.01 | 5 | +| TorchRL | 4 | 3,036.2 ± 24.0 | 32.94 | 2.94 ± 0.02 | 0.73 ± 0.00 | 5 | +| TorchRL | 8 | 3,983.5 ± 25.7 | 25.10 | 3.87 ± 0.03 | 0.48 ± 0.00 | 5 | **Unsupported (recorded, not run)**: +- zeroRL W=2: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=2: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=2: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=2: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=2: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=4: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=4: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=4: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=4: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=4: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=8: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=8: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=8: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=8: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- zeroRL W=8: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported +- SB3 W=2: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=2 unsupported +- SB3 W=2: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=2 unsupported +- SB3 W=2: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=2 unsupported +- SB3 W=2: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=2 unsupported +- SB3 W=2: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=2 unsupported +- SB3 W=4: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=4 unsupported +- SB3 W=4: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=4 unsupported +- SB3 W=4: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=4 unsupported +- SB3 W=4: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=4 unsupported +- SB3 W=4: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=4 unsupported - zeroRL W=2: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported - zeroRL W=2: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported - zeroRL W=2: zeroRL has no multi-process environment path (vectorize_env -> SyncVectorEnv only); W>=2 unsupported @@ -125,115 +121,79 @@ Rows recorded with device=cuda; every statistic in this section is computed on t - SB3 W=4: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=4 unsupported - SB3 W=4: SB3 SubprocVecEnv runs exactly one process per env, so only W=1 (DummyVecEnv) and W=8 (SubprocVecEnv) exist; W=4 unsupported -![throughput](figures/expB_throughput.png) - -![speedup](figures/expB_speedup.png) - -![efficiency](figures/expB_efficiency.png) - - -## Experiment C — PPO training protocol (fixed 2048 steps/env per rollout) - -| framework | N | updates | throughput (trans/s) | mean time (s) | mean return (last 100) | reps | -| --- | --- | --- | --- | --- | --- | --- | -| zeroRL | 1 | 128 | 766.1 ± 4.5 | 342.2 | -156.4 ± 102.0 | 3 | -| zeroRL | 2 | 64 | 967.9 ± 18.2 | 270.9 | -133.3 ± 90.1 | 3 | -| zeroRL | 4 | 32 | 1,274.8 ± 8.4 | 205.6 | -104.6 ± 26.3 | 3 | -| zeroRL | 8 | 16 | 1,283.2 ± 5.5 | 204.3 | -25.3 ± 16.3 | 3 | -| zeroRL | 16 | 8 | 1,453.1 ± 5.5 | 180.4 | -44.6 ± 15.7 | 3 | -| SB3 | 1 | 128 | 723.5 ± 2.6 | 362.3 | -155.6 ± 44.8 | 3 | -| SB3 | 2 | 64 | 877.0 ± 0.8 | 298.9 | -107.0 ± 44.6 | 3 | -| SB3 | 4 | 32 | 1,056.8 ± 4.6 | 248.1 | -104.9 ± 16.5 | 3 | -| SB3 | 8 | 16 | 1,065.9 ± 3.0 | 245.9 | -45.9 ± 9.6 | 3 | -| SB3 | 16 | 8 | 1,186.6 ± 2.6 | 220.9 | -83.8 ± 17.8 | 3 | -| TorchRL | 1 | 128 | 424.2 ± 3.0 | 618.0 | -116.9 ± 49.3 | 3 | -| TorchRL | 2 | 64 | 536.6 ± 2.3 | 488.6 | -107.9 ± 46.7 | 3 | -| TorchRL | 4 | 32 | 689.6 ± 2.1 | 380.1 | 0.2 ± 79.1 | 3 | -| TorchRL | 8 | 16 | 780.2 ± 1.0 | 336.0 | -60.1 ± 27.6 | 3 | -| TorchRL | 16 | 8 | 861.9 ± 3.9 | 304.2 | -47.0 ± 2.1 | 3 | +![throughput](figures/expB_humanoidstandupv5_throughput.png) -![throughput](figures/expC_throughput.png) +![speedup](figures/expB_humanoidstandupv5_speedup.png) -![returns](figures/expC_returns.png) +![efficiency](figures/expB_humanoidstandupv5_efficiency.png) -## Experiment D — PPO training protocol (fixed global rollout of 2048) +## Experiment C — PPO training protocol (fixed 2048 steps/env per rollout) [humanoidstandupv5] | framework | N | updates | throughput (trans/s) | mean time (s) | mean return (last 100) | reps | | --- | --- | --- | --- | --- | --- | --- | -| zeroRL | 1 | 128 | 816.5 ± 4.1 | 321.1 | -156.4 ± 102.0 | 3 | -| zeroRL | 2 | 128 | 1,028.6 ± 1.0 | 254.9 | -93.2 ± 9.3 | 3 | -| zeroRL | 4 | 128 | 1,279.2 ± 3.0 | 204.9 | -162.9 ± 44.9 | 3 | -| zeroRL | 8 | 128 | 1,293.0 ± 5.7 | 202.8 | -59.4 ± 66.5 | 3 | -| zeroRL | 16 | 128 | 1,484.7 ± 5.2 | 176.6 | -68.4 ± 111.2 | 3 | -| SB3 | 1 | 128 | 716.9 ± 1.9 | 365.7 | -155.6 ± 44.8 | 3 | -| SB3 | 2 | 128 | 873.2 ± 3.6 | 300.2 | -53.4 ± 38.9 | 3 | -| SB3 | 4 | 128 | 1,046.4 ± 2.1 | 250.5 | -97.3 ± 112.5 | 3 | -| SB3 | 8 | 128 | 1,063.3 ± 1.6 | 246.5 | -78.1 ± 111.2 | 3 | -| SB3 | 16 | 128 | 1,174.8 ± 5.8 | 223.1 | -96.1 ± 46.7 | 3 | -| TorchRL | 1 | 128 | 431.0 ± 3.0 | 608.2 | -116.9 ± 49.3 | 3 | -| TorchRL | 2 | 128 | 552.3 ± 2.1 | 474.7 | -57.5 ± 80.4 | 3 | -| TorchRL | 4 | 128 | 707.9 ± 2.7 | 370.3 | -117.8 ± 57.2 | 3 | -| TorchRL | 8 | 128 | 803.0 ± 9.3 | 326.5 | -125.5 ± 4.8 | 3 | -| TorchRL | 16 | 128 | 887.8 ± 11.4 | 295.3 | -58.2 ± 98.9 | 3 | - -![throughput](figures/expD_throughput.png) - -![returns](figures/expD_returns.png) +| zeroRL | 1 | 128 | 428.1 ± 2.5 | 612.4 | 75,253.6 ± 5,108.2 | 6 | +| zeroRL | 2 | 64 | 536.7 ± 4.2 | 488.4 | 75,079.7 ± 3,892.0 | 6 | +| zeroRL | 4 | 32 | 650.7 ± 5.7 | 402.9 | 64,966.9 ± 4,545.7 | 6 | +| zeroRL | 8 | 16 | 687.6 ± 7.3 | 381.3 | 57,229.6 ± 1,909.4 | 6 | +| zeroRL | 16 | 8 | 762.7 ± 4.1 | 343.7 | 47,649.0 ± 799.5 | 6 | +| SB3 | 1 | 128 | 418.7 ± 8.6 | 626.3 | 80,410.9 ± 2,362.9 | 6 | +| SB3 | 2 | 64 | 526.3 ± 5.7 | 498.2 | 73,775.3 ± 2,866.8 | 6 | +| SB3 | 4 | 32 | 627.7 ± 5.9 | 417.6 | 70,151.9 ± 1,711.2 | 6 | +| SB3 | 8 | 16 | 655.1 ± 6.0 | 400.2 | 62,676.0 ± 2,365.7 | 6 | +| SB3 | 16 | 8 | 721.4 ± 3.4 | 363.4 | 50,308.9 ± 2,015.7 | 6 | +| TorchRL | 1 | 128 | 264.9 ± 2.9 | 989.6 | 72,593.4 ± 2,624.9 | 6 | +| TorchRL | 2 | 64 | 365.5 ± 3.4 | 717.3 | 73,504.3 ± 3,083.8 | 6 | +| TorchRL | 4 | 32 | 460.4 ± 4.5 | 569.4 | 67,696.1 ± 2,503.8 | 6 | +| TorchRL | 8 | 16 | 503.9 ± 4.2 | 520.3 | 64,621.7 ± 824.4 | 6 | +| TorchRL | 16 | 8 | 560.1 ± 4.2 | 468.1 | 48,612.7 ± 1,557.2 | 6 | + +![throughput](figures/expC_humanoidstandupv5_throughput.png) + +![returns](figures/expC_humanoidstandupv5_returns.png) ## Findings (computed from the data) -- Exp A collection, N=1: fastest = **zeroRL** (1,657 trans/s); also SB3 1,594, TorchRL 826. -- Exp A collection, N=2: fastest = **SB3** (2,449 trans/s); also zeroRL 2,416, TorchRL 1,307. -- Exp A collection, N=4: fastest = **zeroRL** (4,582 trans/s); also SB3 4,577, TorchRL 2,382. -- Exp A collection, N=8: fastest = **zeroRL** (4,755 trans/s); also SB3 4,712, TorchRL 3,376. -- Exp A collection, N=16: fastest = **zeroRL** (8,176 trans/s); also SB3 7,991, TorchRL 5,785. -- Exp B collection, W=1 (N=8): fastest = **zeroRL** (4,474 trans/s); also SB3 4,472, TorchRL 3,309. -- Exp B collection, W=2 (N=8): fastest = **TorchRL** (7,332 trans/s). -- Exp B collection, W=4 (N=8): fastest = **TorchRL** (7,842 trans/s). -- Exp B collection, W=8 (N=8): fastest = **TorchRL** (8,453 trans/s); also SB3 3,966. -- Exp C PPO, N=1: fastest = **zeroRL** (766 trans/s); also SB3 724, TorchRL 424. -- Exp C PPO, N=2: fastest = **zeroRL** (968 trans/s); also SB3 877, TorchRL 537. -- Exp C PPO, N=4: fastest = **zeroRL** (1,275 trans/s); also SB3 1,057, TorchRL 690. -- Exp C PPO, N=8: fastest = **zeroRL** (1,283 trans/s); also SB3 1,066, TorchRL 780. -- Exp C PPO, N=16: fastest = **zeroRL** (1,453 trans/s); also SB3 1,187, TorchRL 862. -- Exp D PPO, N=1: fastest = **zeroRL** (817 trans/s); also SB3 717, TorchRL 431. -- Exp D PPO, N=2: fastest = **zeroRL** (1,029 trans/s); also SB3 873, TorchRL 552. -- Exp D PPO, N=4: fastest = **zeroRL** (1,279 trans/s); also SB3 1,046, TorchRL 708. -- Exp D PPO, N=8: fastest = **zeroRL** (1,293 trans/s); also SB3 1,063, TorchRL 803. -- Exp D PPO, N=16: fastest = **zeroRL** (1,485 trans/s); also SB3 1,175, TorchRL 888. -- zeroRL env-scaling efficiency at N=16: 0.31 ± 0.00 (speedup 4.93 ± 0.05, n=5 paired reps, 16x envs). -- SB3 env-scaling efficiency at N=16: 0.31 ± 0.01 (speedup 5.02 ± 0.13, n=5 paired reps, 16x envs). -- TorchRL env-scaling efficiency at N=16: 0.43 ± 0.02 (speedup 6.94 ± 0.36, n=5 paired reps, 16x envs). +- Exp A collection, N=1: fastest = **SB3** (619 trans/s); also zeroRL 605, TorchRL 369. +- Exp A collection, N=2: fastest = **zeroRL** (807 trans/s); also SB3 792, TorchRL 591. +- Exp A collection, N=4: fastest = **zeroRL** (1,074 trans/s); also SB3 1,059, TorchRL 865. +- Exp A collection, N=8: fastest = **zeroRL** (1,137 trans/s); also SB3 1,117, TorchRL 995. +- Exp A collection, N=16: fastest = **zeroRL** (1,329 trans/s); also SB3 1,304, TorchRL 1,221. +- Exp B collection, W=1 (N=8): fastest = **SB3** (1,168 trans/s); also zeroRL 1,160, TorchRL 1,031. +- Exp B collection, W=2 (N=8): fastest = **TorchRL** (2,102 trans/s). +- Exp B collection, W=4 (N=8): fastest = **TorchRL** (3,036 trans/s). +- Exp B collection, W=8 (N=8): fastest = **TorchRL** (3,984 trans/s); also SB3 2,260. +- Exp C PPO, N=1: fastest = **zeroRL** (428 trans/s); also SB3 419, TorchRL 265. +- Exp C PPO, N=2: fastest = **zeroRL** (537 trans/s); also SB3 526, TorchRL 365. +- Exp C PPO, N=4: fastest = **zeroRL** (651 trans/s); also SB3 628, TorchRL 460. +- Exp C PPO, N=8: fastest = **zeroRL** (688 trans/s); also SB3 655, TorchRL 504. +- Exp C PPO, N=16: fastest = **zeroRL** (763 trans/s); also SB3 721, TorchRL 560. +- zeroRL env-scaling efficiency at N=16: 0.14 ± 0.00 (speedup 2.21 ± 0.01, n=5 paired reps, 16x envs). +- SB3 env-scaling efficiency at N=16: 0.13 ± 0.00 (speedup 2.14 ± 0.01, n=5 paired reps, 16x envs). +- TorchRL env-scaling efficiency at N=16: 0.20 ± 0.00 (speedup 3.23 ± 0.02, n=5 paired reps, 16x envs). - zeroRL worker parallel efficiency at W=1: 1.00 (speedup 1.00, n=5 paired reps, 1x workers). -- SB3 worker parallel efficiency at W=8: 0.11 ± 0.00 (speedup 0.89 ± 0.02, n=5 paired reps, 8x workers). -- TorchRL worker parallel efficiency at W=8: 0.32 ± 0.00 (speedup 2.55 ± 0.04, n=5 paired reps, 8x workers). -- zeroRL N=1: training throughput C=766 vs D=817 trans/s (D faster). -- SB3 N=1: training throughput C=724 vs D=717 trans/s (≈ equal). -- TorchRL N=1: training throughput C=424 vs D=431 trans/s (D faster). +- SB3 worker parallel efficiency at W=8: 0.25 ± 0.01 (speedup 1.97 ± 0.04, n=5 paired reps, 8x workers). +- TorchRL worker parallel efficiency at W=8: 0.48 ± 0.00 (speedup 3.87 ± 0.03, n=5 paired reps, 8x workers). ## Data integrity | experiment | ok | failed | unsupported | | --- | --- | --- | --- | -| A | 162 | 0 | 0 | -| B | 35 | 0 | 25 | -| C | 45 | 0 | 0 | -| D | 45 | 0 | 0 | +| A | 215 | 0 | 0 | +| B | 70 | 0 | 50 | +| C | 90 | 0 | 0 | All ok rows recorded an exact transition budget. -GPU rows: 205 (device=cuda) - -Total rows: 517. +Total rows: 425. ## Appendix — protocol C/D configuration (from a result row) | key | value | | --- | --- | | device | cpu | -| env_id | LunarLander-v3 | +| env_id | HumanoidStandup-v5 | | rollout_steps_per_env | 2048 | | rollout_global | 2048 | | updates | 128 | diff --git a/benchmarks/scalability/results/figures/expA_humanoidstandupv5_efficiency.png b/benchmarks/scalability/results/figures/expA_humanoidstandupv5_efficiency.png new file mode 100644 index 0000000..437fd42 Binary files /dev/null and b/benchmarks/scalability/results/figures/expA_humanoidstandupv5_efficiency.png differ diff --git a/benchmarks/scalability/results/figures/expA_humanoidstandupv5_speedup.png b/benchmarks/scalability/results/figures/expA_humanoidstandupv5_speedup.png new file mode 100644 index 0000000..d3f8fa9 Binary files /dev/null and b/benchmarks/scalability/results/figures/expA_humanoidstandupv5_speedup.png differ diff --git a/benchmarks/scalability/results/figures/expA_humanoidstandupv5_throughput.png b/benchmarks/scalability/results/figures/expA_humanoidstandupv5_throughput.png new file mode 100644 index 0000000..27fceb8 Binary files /dev/null and b/benchmarks/scalability/results/figures/expA_humanoidstandupv5_throughput.png differ diff --git a/benchmarks/scalability/results/figures/expB_humanoidstandupv5_efficiency.png b/benchmarks/scalability/results/figures/expB_humanoidstandupv5_efficiency.png new file mode 100644 index 0000000..fa1ad82 Binary files /dev/null and b/benchmarks/scalability/results/figures/expB_humanoidstandupv5_efficiency.png differ diff --git a/benchmarks/scalability/results/figures/expB_humanoidstandupv5_speedup.png b/benchmarks/scalability/results/figures/expB_humanoidstandupv5_speedup.png new file mode 100644 index 0000000..53dcbe4 Binary files /dev/null and b/benchmarks/scalability/results/figures/expB_humanoidstandupv5_speedup.png differ diff --git a/benchmarks/scalability/results/figures/expB_humanoidstandupv5_throughput.png b/benchmarks/scalability/results/figures/expB_humanoidstandupv5_throughput.png new file mode 100644 index 0000000..2ee0b09 Binary files /dev/null and b/benchmarks/scalability/results/figures/expB_humanoidstandupv5_throughput.png differ diff --git a/benchmarks/scalability/results/figures/expC_humanoidstandupv5_returns.png b/benchmarks/scalability/results/figures/expC_humanoidstandupv5_returns.png new file mode 100644 index 0000000..8a5b6cf Binary files /dev/null and b/benchmarks/scalability/results/figures/expC_humanoidstandupv5_returns.png differ diff --git a/benchmarks/scalability/results/figures/expC_humanoidstandupv5_throughput.png b/benchmarks/scalability/results/figures/expC_humanoidstandupv5_throughput.png new file mode 100644 index 0000000..e542c77 Binary files /dev/null and b/benchmarks/scalability/results/figures/expC_humanoidstandupv5_throughput.png differ diff --git a/benchmarks/scalability/results/figures/expD_throughput.png b/benchmarks/scalability/results/figures/expD_throughput.png index 578a460..1252e6f 100644 Binary files a/benchmarks/scalability/results/figures/expD_throughput.png and b/benchmarks/scalability/results/figures/expD_throughput.png differ diff --git a/benchmarks/scalability/results_schema.py b/benchmarks/scalability/results_schema.py index 4570cf8..0ea8a5b 100644 --- a/benchmarks/scalability/results_schema.py +++ b/benchmarks/scalability/results_schema.py @@ -10,6 +10,7 @@ import json import platform +import re import subprocess import sys from datetime import datetime, timezone @@ -43,6 +44,13 @@ ) +def env_slug(env_id: str) -> str: + """Filesystem-safe slug for an env id (e.g. ``HumanoidStandup-v4`` -> + ``humanoidstandupv4``) — used in result/figure filenames so per-env + campaigns can never collide with the default-env files.""" + return re.sub(r"[^a-z0-9]", "", env_id.lower()) + + def environment_info(device: str = "cpu") -> dict[str, Any]: """Capture versions/hardware metadata recorded into every row. diff --git a/benchmarks/scalability/run.py b/benchmarks/scalability/run.py index bbd30fd..a5fcc9d 100644 --- a/benchmarks/scalability/run.py +++ b/benchmarks/scalability/run.py @@ -47,13 +47,19 @@ PPOProtocolConfig, run_ppo_training, ) -from benchmarks.scalability.results_schema import append_row, make_row # noqa: E402 +from benchmarks.scalability.results_schema import ( # noqa: E402 + append_row, + env_slug, + make_row, +) from benchmarks.scalability.workers import ( # noqa: E402 NUM_ENVS_B, UnsupportedConfiguration, run_b_collection, ) +DEFAULT_ENV = "LunarLander-v3" + def _parse_ints(text: str) -> list[int]: return [int(x) for x in text.split(",") if x.strip()] @@ -90,9 +96,13 @@ def build_parser() -> argparse.ArgumentParser: p.add_argument("--measure", type=int, default=100_000) p.add_argument("--total", type=int, default=DEFAULT_PPO_TOTAL, help="total PPO transitions for experiments C/D") + p.add_argument("--env", default=DEFAULT_ENV, + help=f"gymnasium env id (default {DEFAULT_ENV}; a " + "non-default env writes to exp_.jsonl)") p.add_argument("--out", default=None, help="JSONL output (default: benchmarks/scalability/results/" - "exp.jsonl)") + "exp.jsonl, with _ appended for " + "non-default --env)") p.add_argument("--device", choices=["cpu", "cuda"], default="cpu", help="compute device for policies/agents (default cpu; " "env stepping is always CPU)") @@ -104,13 +114,18 @@ def main(argv: list[str] | None = None) -> int: frameworks = _parse_frameworks(args.frameworks) is_ppo = args.experiment in EXPERIMENTS_CD reps = args.reps if args.reps is not None else (3 if is_ppo else 5) - out = Path(args.out) if args.out else ( - REPO_ROOT / "benchmarks/scalability" / "results" / f"exp{args.experiment}.jsonl" - ) + if args.out: + out = Path(args.out) + else: + sfx = f"_{env_slug(args.env)}" if args.env != DEFAULT_ENV else "" + out = (REPO_ROOT / "benchmarks/scalability" / "results" + / f"exp{args.experiment}{sfx}.jsonl") out.parent.mkdir(parents=True, exist_ok=True) + print(f"[env] {args.env} -> {out}", flush=True) ppo_cfg: PPOProtocolConfig | None = None - cfg = CollectionConfig(warmup=args.warmup, measure=args.measure) + cfg = CollectionConfig(env_id=args.env, warmup=args.warmup, + measure=args.measure) # (framework, num_envs, num_workers, mode) if args.experiment == "A": @@ -127,7 +142,7 @@ def main(argv: list[str] | None = None) -> int: for w in workers ] else: - ppo_cfg = PPOProtocolConfig(total=args.total) + ppo_cfg = PPOProtocolConfig(env_id=args.env, total=args.total) env_counts = _parse_ints(args.envs) for n in env_counts: try: diff --git a/benchmarks/scalability/workers.py b/benchmarks/scalability/workers.py index bcfa15a..e536fd6 100644 --- a/benchmarks/scalability/workers.py +++ b/benchmarks/scalability/workers.py @@ -34,8 +34,11 @@ run_zerorl_collection, ) from benchmarks.scalability.policy_reference import ( + ACT_DIM, + OBS_DIM, TorchRLSharedTrunk, build_reference_state_dict, + env_dims, ) NUM_ENVS_B = 8 @@ -117,7 +120,15 @@ def run_torchrl_mp_collection(cfg: CollectionConfig, num_envs: int, with lifecycle() as lc: t0 = time.perf_counter() - trunk = TorchRLSharedTrunk(build_reference_state_dict(cfg.reference_seed)) + d = env_dims(cfg.env_id) + sd = build_reference_state_dict(cfg.reference_seed, obs_dim=d.obs_n, + act_dim=d.act_n, + is_discrete=d.is_discrete) + # default (LunarLander) path keeps the ORIGINAL call signature + if (d.obs_n, d.act_n, d.is_discrete) == (OBS_DIM, ACT_DIM, True): + trunk = TorchRLSharedTrunk(sd) + else: + trunk = TorchRLSharedTrunk(sd, obs_dim=d.obs_n, act_dim=d.act_n) policy = _TorchRLCollectPolicy(trunk) tdm = TensorDictModule(policy, in_keys=["observation"], out_keys=["action", "log_prob", "entropy", "value"]) @@ -145,7 +156,7 @@ def run_torchrl_mp_collection(cfg: CollectionConfig, num_envs: int, first_obs = [round(float(x), 6) for x in td["observation"][0, 0].reshape(-1)[:4]] _absorb_actions(fb, td["action"].movedim(-1, 0).reshape(-1).tolist(), - warm_tracker) + warm_tracker, is_discrete=d.is_discrete) warm_tracker.record(n) warm_s = time.perf_counter() - t _verify_transition_count(warm_tracker.actual, cfg.warmup, "warmup") @@ -170,7 +181,7 @@ def run_torchrl_mp_collection(cfg: CollectionConfig, num_envs: int, for td in collector2: n = int(td.batch_size.numel()) _absorb_actions(fb, td["action"].movedim(-1, 0).reshape(-1).tolist(), - meas_tracker) + meas_tracker, is_discrete=d.is_discrete) meas_tracker.record(n) meas_s = time.perf_counter() - t _verify_transition_count(meas_tracker.actual, cfg.measure, "measure") diff --git a/tests/env/conftest.py b/tests/env/conftest.py index c5719e0..51de628 100644 --- a/tests/env/conftest.py +++ b/tests/env/conftest.py @@ -1,21 +1,48 @@ """Shared fixtures for env tests.""" import os +import subprocess import sys +from functools import lru_cache import pytest +# Minimal scene + the exact operation that has crashed CI natively: +# creating a mujoco.Renderer (EGL/OSMesa context). A native SIGABRT inside +# it would kill the whole pytest session, so the capability check runs in a +# SUBPROCESS where a crash/timeout can only fail the probe, not the suite. +_RENDER_PROBE = """ +import mujoco +m = mujoco.MjModel.from_xml_string( + "" +) +mujoco.Renderer(m, 32, 32) +""" -def _mujoco_gl_unavailable() -> bool: - backend = os.environ.get("MUJOCO_GL", "").lower() - if backend in {"osmesa", "egl"}: + +@lru_cache(maxsize=1) +def mujoco_render_works() -> bool: + """True only if a real MuJoCo renderer can be created here. + + Trusts nothing (env vars like MUJOCO_GL are inherited by the probe but + are claims, not capabilities): the probe either exits 0, times out, or + crashes — anything but 0 means rendering is unavailable on this host. + """ + try: + proc = subprocess.run( + [sys.executable, "-c", _RENDER_PROBE], + capture_output=True, text=True, timeout=30, + env=dict(os.environ), + ) + return proc.returncode == 0 + except Exception: return False - return sys.platform.startswith("linux") and not os.environ.get("DISPLAY") @pytest.fixture def mujoco_gl() -> None: - if _mujoco_gl_unavailable(): + if not mujoco_render_works(): pytest.skip( - "MuJoCo OpenGL renderer unavailable (headless; set MUJOCO_GL=osmesa|egl)" + "MuJoCo renderer unavailable on this host " + "(no usable GL/EGL context; probe failed or crashed)" ) diff --git a/tests/env/test_hopper_immediate.py b/tests/env/test_hopper_immediate.py index ad85a41..ab03af1 100644 --- a/tests/env/test_hopper_immediate.py +++ b/tests/env/test_hopper_immediate.py @@ -1,5 +1,14 @@ """Tests for Hopper2D (MujocoEnv) and its immediate-mode training loop.""" + +import pytest as _pytest + +try: # headless CI hosts may lack EGL: mujoco's import chain then dies in + # PyOpenGL with AttributeError, which would abort pytest COLLECTION + import mujoco # noqa: F401 +except Exception: # pragma: no cover - host-dependent + _pytest.skip("mujoco (or its OpenGL/EGL stack) unavailable on this host", + allow_module_level=True) import numpy as np import pytest import torch diff --git a/tests/env/test_mujoco.py b/tests/env/test_mujoco.py index b6cabcb..8bd51b3 100644 --- a/tests/env/test_mujoco.py +++ b/tests/env/test_mujoco.py @@ -1,5 +1,14 @@ """Unit tests for MujocoEnv.""" + +import pytest as _pytest + +try: # headless CI hosts may lack EGL: mujoco's import chain then dies in + # PyOpenGL with AttributeError, which would abort pytest COLLECTION + import mujoco # noqa: F401 +except Exception: # pragma: no cover - host-dependent + _pytest.skip("mujoco (or its OpenGL/EGL stack) unavailable on this host", + allow_module_level=True) from pathlib import Path import gymnasium as gym diff --git a/tests/test_benchmark_scaling.py b/tests/test_benchmark_scaling.py index 63c4ca6..19f605f 100644 --- a/tests/test_benchmark_scaling.py +++ b/tests/test_benchmark_scaling.py @@ -1694,3 +1694,125 @@ def boom(*_args, **_kwargs): assert row["status"] == "failed" assert "cli-boom-marker" in row["error"] assert "Traceback" in row["error"] + + +class TestEnvAgnosticAdaptation: + """Env-agnostic + continuous-action adaptation of the shared reference + policy (HumanoidStandup-v4 as the continuous case).""" + + def test_env_slug_and_dims(self) -> None: + assert schema.env_slug("HumanoidStandup-v4") == "humanoidstandupv4" + assert schema.env_slug("LunarLander-v3") == "lunarlanderv3" + + d = ref.env_dims("LunarLander-v3") + assert (d.obs_n, d.act_n, d.is_discrete) == (8, 4, True) + + d = ref.env_dims("HumanoidStandup-v4") # gym.make; lru-cached + assert d.obs_shape == (376,) + assert d.act_n == 17 + assert d.is_discrete is False + + def test_humanoid_reference_sd_shapes(self) -> None: + sd = ref.build_reference_state_dict(4242, obs_dim=376, act_dim=17, + is_discrete=False) + assert sd["extract_layer.0.weight"].shape == (64, 376) + assert sd["actor.weight"].shape == (17, 64) + assert "log_std" in sd and sd["log_std"].shape == (17,) + + trunk = ref.TorchRLSharedTrunk(sd, obs_dim=376, act_dim=17) + assert isinstance(trunk.log_std, torch.nn.Parameter) + assert any(p is trunk.log_std for p in trunk.parameters()) + + # default (LunarLander-shaped) sd is unchanged by the adaptation + default_sd = ref.build_reference_state_dict() + assert default_sd["extract_layer.0.weight"].shape == (64, 8) + assert "log_std" not in default_sd + + def test_continuous_cross_framework_equivalence(self) -> None: + pytest.importorskip("stable_baselines3") + pytest.importorskip("torchrl") + import gymnasium as gym + + from benchmarks.scalability.collection import _TorchRLCollectPolicy + + sd = ref.build_reference_state_dict(4242, obs_dim=376, act_dim=17, + is_discrete=False) + agent = ref.build_deterministic_zerorl_agent( + sd, obs_dim=376, act_dim=17, is_discrete=False) + sb3 = ref.build_sb3_shared_policy( + gym.spaces.Box(-1.0, 1.0, shape=(376,), dtype="float32"), + gym.spaces.Box(-1.0, 1.0, shape=(17,), dtype="float32")) + ref.load_into_sb3(sb3, sd) + trunk = ref.TorchRLSharedTrunk(sd, obs_dim=376, act_dim=17) + torchrl_policy = _TorchRLCollectPolicy(trunk) + + torch.manual_seed(0) + obs = torch.randn(8, 376) + + z = agent.get_action(obs) + s_actions, s_values, s_lp, s_ent = sb3.forward_collection(obs) + t_actions, t_lp, t_ent, t_value = torchrl_policy(obs) + + for label, actions, values, lp, ent in ( + ("SB3", s_actions, s_values.squeeze(-1), s_lp, s_ent), + ("TorchRL", t_actions, t_value, t_lp, t_ent), + ): + assert torch.allclose(actions, z["action"], atol=1e-5), label + assert torch.allclose(values, z["value"], atol=1e-5), label + assert torch.allclose(lp, z["log_prob"], atol=1e-5), label + assert torch.allclose(ent, z["entropy"], atol=1e-5), label + + def test_env_flag_and_auto_out(self, monkeypatch, tmp_path) -> None: + from pathlib import Path + + from benchmarks.resources import report as resources_report + from benchmarks.resources import plots as resources_plots + from benchmarks.scalability import plots as scaling_plots + from benchmarks.scalability import report as scaling_report + + run_mod = __import__("benchmarks.scalability.run", fromlist=["run"]) + res_run = __import__("benchmarks.resources.run", fromlist=["run"]) + + # run.py: --env default + override (slug flag lives on report/plots) + for mod in (run_mod, res_run): + assert mod.build_parser().parse_args([]).env == "LunarLander-v3" + parsed = mod.build_parser().parse_args( + ["--env", "HumanoidStandup-v4"]) + assert parsed.env == "HumanoidStandup-v4" + + # resources report/plots expose build_parser: --env-slug default "" + # + accepted override + for mod in (resources_report, resources_plots): + assert mod.build_parser().parse_args([]).env_slug == "" + parsed = mod.build_parser().parse_args(["--env-slug", "x"]) + assert parsed.env_slug == "x" + + # scalability report/plots build their parser inline in main(): + # capture the slug actually forwarded to the generator + results_dir = tmp_path / "results" + results_dir.mkdir() + common = ["--results-dir", str(results_dir), + "--out", str(tmp_path / "out")] + for mod, fn_name, stub in ( + (scaling_report, "generate_report", + lambda: tmp_path / "REPORT.md"), + (scaling_plots, "generate_all", + lambda: [Path("f.png")]), + ): + captured: list[str] = [] + + # avoid touching real results: replace, then only capture slug + def stub_fn(*_args, __cap=captured, **kwargs): + __cap.append(kwargs.get("slug", "")) + return stub() + + monkeypatch.setattr(mod, fn_name, stub_fn) + assert mod.main(common) == 0 + assert mod.main(common + ["--env-slug", "x"]) == 0 + assert captured == ["", "x"], (fn_name, captured) + + # auto-out composition used by run.py mains for a non-default env + assert ("expA_" + schema.env_slug("HumanoidStandup-v4") + + ".jsonl") == "expA_humanoidstandupv4.jsonl" + assert ("resources_" + schema.env_slug("HumanoidStandup-v4") + + ".jsonl") == "resources_humanoidstandupv4.jsonl" diff --git a/tests/test_resources.py b/tests/test_resources.py index a44ea8f..6d5218f 100644 --- a/tests/test_resources.py +++ b/tests/test_resources.py @@ -666,3 +666,16 @@ def test_cli_parser_defaults() -> None: assert args.workload == "all" assert args.interval == DEFAULT_INTERVAL_S assert args.reps_collect == 5 and args.reps_ppo == 3 + + +def test_resources_env_flag_and_slug() -> None: + from benchmarks.resources.run import build_parser + from benchmarks.scalability.results_schema import env_slug + + args = build_parser().parse_args([]) + assert args.env == "LunarLander-v3" + args = build_parser().parse_args(["--env", "HumanoidStandup-v4"]) + assert args.env == "HumanoidStandup-v4" + # non-default env => slug-prefixed sample files (run.py composition) + assert f"{env_slug('HumanoidStandup-v4')}_collect_zeroRL_rep0.jsonl" == \ + "humanoidstandupv4_collect_zeroRL_rep0.jsonl"