diff --git a/benchmarks/benchmarks/scalability/results/expA_gpu.jsonl b/benchmarks/benchmarks/scalability/results/expA_gpu.jsonl deleted file mode 100644 index 23f339f..0000000 --- a/benchmarks/benchmarks/scalability/results/expA_gpu.jsonl +++ /dev/null @@ -1,134 +0,0 @@ -{"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 deleted file mode 100644 index 283d4f3..0000000 --- a/benchmarks/exp_A_gpu.jsonl +++ /dev/null @@ -1,75 +0,0 @@ -{"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/results/REPORT.md b/benchmarks/resources/results/REPORT.md index be665b8..8342d71 100644 --- a/benchmarks/resources/results/REPORT.md +++ b/benchmarks/resources/results/REPORT.md @@ -1,6 +1,6 @@ # Resource-consumption benchmark report -Generated: 2026-11-03T19:43:24.430693+00:00 · results: /home/darrius/project/zerorl/benchmarks/resources/results +Generated: 2026-11-08T23:35:45.924829+00:00 · results: /home/darrius/project/zerorl/benchmarks/resources/results ## Resource consumption (experiment R) diff --git a/benchmarks/resources/results/REPORT_humanoidstandupv4.md b/benchmarks/resources/results/REPORT_humanoidstandupv4.md new file mode 100644 index 0000000..eb8c8af --- /dev/null +++ b/benchmarks/resources/results/REPORT_humanoidstandupv4.md @@ -0,0 +1,58 @@ +# Resource-consumption benchmark report + +Generated: 2026-11-08T23:35:46.556160+00:00 · results: /home/darrius/project/zerorl/benchmarks/resources/results + +## Resource consumption (experiment R) [humanoidstandupv4] + +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 the raw rows); GPU = device-level telemetry only (workload does not use the GPU); energy recorded only when reliable telemetry exists. + +| workload | framework | wall (s) | cpu util avg (%) | cpu total (s) | cpu s / 1k | rss peak (MiB) | n | +| --- | --- | --- | --- | --- | --- | --- | --- | +| collect | zeroRL | 112.0 ± 5.2 | 223.0 ± 1.8 | 249.2 ± 9.9 | 2.49 ± 0.10 | 1,269.0 ± 170.4 | 5 | +| collect | SB3 | 105.6 ± 7.4 | 226.7 ± 2.5 | 239.3 ± 13.9 | 2.39 ± 0.14 | 1,632.6 ± 74.2 | 5 | +| collect | TorchRL | 121.9 ± 3.1 | 207.1 ± 1.2 | 251.6 ± 4.0 | 2.52 ± 0.04 | 1,307.8 ± 57.5 | 5 | +| ppo | zeroRL | 421.0 ± 29.7 | 395.7 ± 18.6 | 1,672.8 ± 39.7 | 6.38 ± 0.15 | 1,441.2 ± 270.0 | 3 | +| ppo | SB3 | 416.4 ± 16.1 | 432.5 ± 5.0 | 1,806.2 ± 52.2 | 6.89 ± 0.20 | 1,295.9 ± 9.8 | 3 | +| ppo | TorchRL | 670.3 ± 24.2 | 385.3 ± 3.5 | 2,580.7 ± 88.4 | 9.84 ± 0.34 | 1,335.1 ± 17.1 | 3 | + +- GPU telemetry: available (device-level only). +- Energy: unavailable — GPU power telemetry unsupported on this device and no RAPL/powercap energy interface for CPU/whole-system energy. +- Raw per-run rows: `results/resources_humanoidstandupv4.jsonl` (validation runs kept separately in `resources_validation.*`). + +_Memory note: raw values are bytes (v2; v1 rows/sample files used KiB); averages are time-weighted (trapezoidal); summed RSS may double-count shared pages — PSS/USS are the attributable views; v1 rows show — (never 0)._ + +### Memory (RSS / PSS / USS) — collect + +**Peak (end-to-end)** + +| framework | rss (MiB) | pss (MiB) | uss (MiB) | peak rss — measure phase (MiB) | n | +| --- | --- | --- | --- | --- | --- | +| zeroRL | 1,269.0 ± 170.4 | 1,266.4 ± 170.5 | 1,265.7 ± 170.5 | 1,197.7 ± 121.0 | 5 | +| SB3 | 1,632.6 ± 74.2 | 1,630.4 ± 74.2 | 1,629.6 ± 74.2 | 1,632.6 ± 74.2 | 5 | +| TorchRL | 1,307.8 ± 57.5 | 1,305.6 ± 57.6 | 1,304.8 ± 57.6 | 1,305.3 ± 62.4 | 5 | + +**Average (end-to-end, time-weighted)** + +| framework | rss (MiB) | pss (MiB) | uss (MiB) | n | +| --- | --- | --- | --- | --- | +| zeroRL | 1,203.0 ± 119.9 | 1,200.7 ± 119.8 | 1,199.9 ± 119.8 | 5 | +| SB3 | 1,475.3 ± 64.7 | 1,473.2 ± 64.5 | 1,472.4 ± 64.5 | 5 | +| TorchRL | 1,285.9 ± 59.3 | 1,283.7 ± 59.2 | 1,282.9 ± 59.2 | 5 | + +### Memory (RSS / PSS / USS) — ppo + +**Peak (end-to-end)** + +| framework | rss (MiB) | pss (MiB) | uss (MiB) | peak rss — measure phase (MiB) | n | +| --- | --- | --- | --- | --- | --- | +| zeroRL | 1,441.2 ± 270.0 | 1,376.2 ± 175.7 | 1,316.9 ± 86.2 | 1,441.2 ± 270.0 | 3 | +| SB3 | 1,295.9 ± 9.8 | 1,285.2 ± 9.8 | 1,277.6 ± 9.8 | 1,295.8 ± 9.8 | 3 | +| TorchRL | 1,335.1 ± 17.1 | 1,324.9 ± 16.7 | 1,317.3 ± 16.7 | 1,321.3 ± 21.1 | 3 | + +**Average (end-to-end, time-weighted)** + +| framework | rss (MiB) | pss (MiB) | uss (MiB) | n | +| --- | --- | --- | --- | --- | +| zeroRL | 1,284.7 ± 1.2 | 1,273.9 ± 1.6 | 1,266.0 ± 2.0 | 3 | +| SB3 | 1,295.8 ± 9.8 | 1,285.2 ± 9.8 | 1,277.6 ± 9.8 | 3 | +| TorchRL | 1,321.2 ± 21.0 | 1,311.0 ± 21.0 | 1,303.4 ± 21.0 | 3 | diff --git a/benchmarks/scalability/collection.py b/benchmarks/scalability/collection.py index 5476312..51d16dd 100644 --- a/benchmarks/scalability/collection.py +++ b/benchmarks/scalability/collection.py @@ -188,7 +188,10 @@ def _absorb_actions(fb: dict[str, Any], actions: list[int], if len(fb["actions_head"]) < FINGERPRINT_HEAD: fb["actions_head"].append(int(a) if is_discrete else round(float(a), 6)) - total = int(sum(actions)) + # int actions sum to int (unchanged); continuous sums MUST stay float — + # int() truncation made every Box action sum to 0 and broke parity with + # zeroRL's float(np.sum(buffer_actions)) fingerprint. + total = sum(actions) if tracker.target == fb.get("_warm_target"): fb["action_sum_warmup"] += total else: @@ -381,14 +384,22 @@ def run_zerorl_collection(cfg: CollectionConfig, num_envs: int, seed: int, # fingerprints from the (official-path) buffer data = buffer.get_all() - actions = data["action"].reshape(-1).tolist() + flat = data["action"].reshape(-1) + # action elements per vectorized step: 1 for scalar (discrete) + # actions, act_dim for Box actions — the warmup slice must cover + # whole transitions, matching what _absorb_actions accumulates for + # SB3/TorchRL (all warmup action values), not a flat-element prefix. + act_dim = flat.numel() // (data["action"].shape[0] + * data["action"].shape[1]) + actions = flat.tolist() warm_n = cfg.warmup + warm_elems = warm_n * act_dim fb = { "first_obs_head": [round(float(x), 6) for x in first_obs.reshape(-1)[:4]], "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:])), + "action_sum_warmup": float(np.sum(actions[:warm_elems])), + "action_sum_measure": float(np.sum(actions[warm_elems:])), } warm_t = BudgetTracker(cfg.warmup) @@ -616,7 +627,7 @@ def run_torchrl_collection(cfg: CollectionConfig, num_envs: int, seed: int, for x in td["observation"][0].reshape(-1)[:4]] # 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(), + _absorb_actions(fb, td["action"].movedim(1, 0).reshape(-1).tolist(), warm_tracker, is_discrete=d.is_discrete) warm_tracker.record(n) warm_s = time.perf_counter() - t @@ -640,7 +651,7 @@ def run_torchrl_collection(cfg: CollectionConfig, num_envs: int, seed: int, t = time.perf_counter() for td in collector2: n = int(td.batch_size.numel()) - _absorb_actions(fb, td["action"].movedim(-1, 0).reshape(-1).tolist(), + _absorb_actions(fb, td["action"].movedim(1, 0).reshape(-1).tolist(), meas_tracker, is_discrete=d.is_discrete) for key in meas_fields: meas_fields[key] += n diff --git a/benchmarks/scalability/policy_reference.py b/benchmarks/scalability/policy_reference.py index 73c8798..fecfe4c 100644 --- a/benchmarks/scalability/policy_reference.py +++ b/benchmarks/scalability/policy_reference.py @@ -68,10 +68,12 @@ class EnvDims(NamedTuple): 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). + Discrete -> ``act_shape=()``, ``act_n=space.n``; Box -> ``act_shape= + space.shape``, ``act_n=int(np.prod(shape))``. ``obs_n`` is the FLAT + integer input size for the benchmark MLPs — ``int(np.prod(obs_shape))`` + for every rank (``get_obs_act`` now returns the observation *shape* + tuple as ``obs_n``; benchmark trunks stay MLP-only, so flat ints are + the correct contract here). """ env = gym.make(env_id) try: @@ -86,7 +88,7 @@ def env_dims(env_id: str) -> EnvDims: 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] + obs_n = int(np.prod(obs_shape)) return EnvDims(obs_shape, act_shape, obs_n, act_n, is_discrete) diff --git a/benchmarks/scalability/results/REPORT.md b/benchmarks/scalability/results/REPORT.md deleted file mode 100644 index 19418f8..0000000 --- a/benchmarks/scalability/results/REPORT.md +++ /dev/null @@ -1,217 +0,0 @@ -# Scalability benchmark report - -Generated: 2026-11-07T19:17:21.500812+00:00 · rows: 425 · figures in `figures/` - -## Environment - -| key | value | -| --- | --- | -| python | 3.13.14 | -| platform | Linux-6.8.12-tegra-aarch64-with-glibc2.39 | -| processor | aarch64 | -| device | cpu | -| 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 | - -## Methodology (summary) - -- Frameworks: zeroRL (this repo, official `rollout_phase` / `BaseTrain.train()`), Stable-Baselines3, TorchRL — same shared-trunk policy/weights where policies apply (see `policy_reference.py`). -- Experiments A/B count individual env transitions with a warmup of 16,000 (excluded) + 100,000 measured transitions per run; Experiments C/D train PPO for exactly 262,144 transitions per run. -- Repetitions: 5 seeds per cell for A/B, 3 for C/D; every reported statistic aggregates **all** ok repetitions (mean ± std), never best-of. -- Metrics: `throughput = measured / measure_seconds`, `speedup(N) = t(baseline) / t(N)`, `environment_scaling_efficiency(N) = speedup(N) / N`, `worker_parallel_efficiency(W) = speedup(W) / (W / 1)`. -- 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) [humanoidstandupv5] - -| framework | N | throughput (trans/s) | mean time (s) | speedup (paired) | env-scaling efficiency (paired) | pairs (n) | -| --- | --- | --- | --- | --- | --- | --- | -| 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_humanoidstandupv5_throughput.png) - -![speedup](figures/expA_humanoidstandupv5_speedup.png) - -![efficiency](figures/expA_humanoidstandupv5_efficiency.png) - - -## 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 | 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 -- 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 - -![throughput](figures/expB_humanoidstandupv5_throughput.png) - -![speedup](figures/expB_humanoidstandupv5_speedup.png) - -![efficiency](figures/expB_humanoidstandupv5_efficiency.png) - - -## 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 | 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 = **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: 0.25 ± 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 | 215 | 0 | 0 | -| B | 70 | 0 | 50 | -| C | 90 | 0 | 0 | - -All ok rows recorded an exact transition budget. - -Total rows: 425. - -## Appendix — protocol C/D configuration (from a result row) - -| key | value | -| --- | --- | -| device | cpu | -| env_id | HumanoidStandup-v5 | -| rollout_steps_per_env | 2048 | -| rollout_global | 2048 | -| updates | 128 | -| total | 262144 | -| lr | 0.0003 | -| gamma | 0.99 | -| gae_lambda | 0.95 | -| clip_eps | 0.2 | -| ent_coef | 0.01 | -| vf_coef | 0.5 | -| epochs | 10 | -| batch_size | 64 | -| grad_clip | 0.5 | -| lr_start | 0.0003 | -| lr_end | 0.0 | -| optimizer | {'type': 'Adam', 'eps': 1e-05, 'schedule': 'linear_decay_to_zero'} | -| net_arch | [64, 64] | -| reference_seed | 4242 | - ---- -See `README.md` (reproduction), `INSPECTION.md` (framework findings), `results/*.jsonl` (raw data). diff --git a/benchmarks/scalability/results/figures/expA_efficiency.png b/benchmarks/scalability/results/figures/expA_efficiency.png deleted file mode 100644 index 692bfc8..0000000 Binary files a/benchmarks/scalability/results/figures/expA_efficiency.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expA_humanoidstandupv5_efficiency.png b/benchmarks/scalability/results/figures/expA_humanoidstandupv5_efficiency.png deleted file mode 100644 index 437fd42..0000000 Binary files a/benchmarks/scalability/results/figures/expA_humanoidstandupv5_efficiency.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expA_humanoidstandupv5_speedup.png b/benchmarks/scalability/results/figures/expA_humanoidstandupv5_speedup.png deleted file mode 100644 index d3f8fa9..0000000 Binary files a/benchmarks/scalability/results/figures/expA_humanoidstandupv5_speedup.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expA_humanoidstandupv5_throughput.png b/benchmarks/scalability/results/figures/expA_humanoidstandupv5_throughput.png deleted file mode 100644 index 27fceb8..0000000 Binary files a/benchmarks/scalability/results/figures/expA_humanoidstandupv5_throughput.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expA_speedup.png b/benchmarks/scalability/results/figures/expA_speedup.png deleted file mode 100644 index 238aec7..0000000 Binary files a/benchmarks/scalability/results/figures/expA_speedup.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expA_throughput.png b/benchmarks/scalability/results/figures/expA_throughput.png deleted file mode 100644 index 9f1dacd..0000000 Binary files a/benchmarks/scalability/results/figures/expA_throughput.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expB_efficiency.png b/benchmarks/scalability/results/figures/expB_efficiency.png deleted file mode 100644 index a3c37d7..0000000 Binary files a/benchmarks/scalability/results/figures/expB_efficiency.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expB_humanoidstandupv5_efficiency.png b/benchmarks/scalability/results/figures/expB_humanoidstandupv5_efficiency.png deleted file mode 100644 index fa1ad82..0000000 Binary files a/benchmarks/scalability/results/figures/expB_humanoidstandupv5_efficiency.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expB_humanoidstandupv5_speedup.png b/benchmarks/scalability/results/figures/expB_humanoidstandupv5_speedup.png deleted file mode 100644 index 53dcbe4..0000000 Binary files a/benchmarks/scalability/results/figures/expB_humanoidstandupv5_speedup.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expB_humanoidstandupv5_throughput.png b/benchmarks/scalability/results/figures/expB_humanoidstandupv5_throughput.png deleted file mode 100644 index 2ee0b09..0000000 Binary files a/benchmarks/scalability/results/figures/expB_humanoidstandupv5_throughput.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expB_speedup.png b/benchmarks/scalability/results/figures/expB_speedup.png deleted file mode 100644 index c157559..0000000 Binary files a/benchmarks/scalability/results/figures/expB_speedup.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expB_throughput.png b/benchmarks/scalability/results/figures/expB_throughput.png deleted file mode 100644 index 9327293..0000000 Binary files a/benchmarks/scalability/results/figures/expB_throughput.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expC_humanoidstandupv5_returns.png b/benchmarks/scalability/results/figures/expC_humanoidstandupv5_returns.png deleted file mode 100644 index 8a5b6cf..0000000 Binary files a/benchmarks/scalability/results/figures/expC_humanoidstandupv5_returns.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expC_humanoidstandupv5_throughput.png b/benchmarks/scalability/results/figures/expC_humanoidstandupv5_throughput.png deleted file mode 100644 index e542c77..0000000 Binary files a/benchmarks/scalability/results/figures/expC_humanoidstandupv5_throughput.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expC_returns.png b/benchmarks/scalability/results/figures/expC_returns.png deleted file mode 100644 index 07bd390..0000000 Binary files a/benchmarks/scalability/results/figures/expC_returns.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expC_throughput.png b/benchmarks/scalability/results/figures/expC_throughput.png deleted file mode 100644 index 8b967db..0000000 Binary files a/benchmarks/scalability/results/figures/expC_throughput.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expD_returns.png b/benchmarks/scalability/results/figures/expD_returns.png deleted file mode 100644 index a7666f4..0000000 Binary files a/benchmarks/scalability/results/figures/expD_returns.png and /dev/null differ diff --git a/benchmarks/scalability/results/figures/expD_throughput.png b/benchmarks/scalability/results/figures/expD_throughput.png deleted file mode 100644 index 1252e6f..0000000 Binary files a/benchmarks/scalability/results/figures/expD_throughput.png and /dev/null differ diff --git a/benchmarks/scalability/run.py b/benchmarks/scalability/run.py index a5fcc9d..63f9d06 100644 --- a/benchmarks/scalability/run.py +++ b/benchmarks/scalability/run.py @@ -92,6 +92,10 @@ def build_parser() -> argparse.ArgumentParser: help="repetitions (default: 5 for A/B, 3 for C/D)") p.add_argument("--seed", type=int, default=1000, help="base seed; rep r runs with seed+r (recorded per row)") + p.add_argument("--rep-start", type=int, default=0, + help="first repetition index (rows get rep=rep-start..; " + "lets memory-constrained hosts run reps in separate " + "fresh processes without duplicating rep indices)") p.add_argument("--warmup", type=int, default=16_000) p.add_argument("--measure", type=int, default=100_000) p.add_argument("--total", type=int, default=DEFAULT_PPO_TOTAL, @@ -153,7 +157,7 @@ def main(argv: list[str] | None = None) -> int: n_rows = n_failed = n_unsupported = 0 for framework, num_envs, num_workers, mode in specs: - for rep in range(reps): + for rep in range(args.rep_start, args.rep_start + reps): seed = args.seed + rep label = (f"{framework:>7} N={num_envs:>2} W={num_workers} " f"rep={rep} seed={seed}") diff --git a/benchmarks/scalability/workers.py b/benchmarks/scalability/workers.py index e536fd6..80c2a7a 100644 --- a/benchmarks/scalability/workers.py +++ b/benchmarks/scalability/workers.py @@ -155,7 +155,7 @@ def run_torchrl_mp_collection(cfg: CollectionConfig, num_envs: int, if first_obs is None: 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(), + _absorb_actions(fb, td["action"].movedim(1, 0).reshape(-1).tolist(), warm_tracker, is_discrete=d.is_discrete) warm_tracker.record(n) warm_s = time.perf_counter() - t @@ -180,7 +180,7 @@ def run_torchrl_mp_collection(cfg: CollectionConfig, num_envs: int, t = time.perf_counter() for td in collector2: n = int(td.batch_size.numel()) - _absorb_actions(fb, td["action"].movedim(-1, 0).reshape(-1).tolist(), + _absorb_actions(fb, td["action"].movedim(1, 0).reshape(-1).tolist(), meas_tracker, is_discrete=d.is_discrete) meas_tracker.record(n) meas_s = time.perf_counter() - t diff --git a/examples/air_raid.py b/examples/air_raid.py new file mode 100644 index 0000000..ca6152d --- /dev/null +++ b/examples/air_raid.py @@ -0,0 +1,16 @@ +import torch +from zerorl.algorithms.ppo import easy_train_ppo +from zerorl.config import AlgoConfig, TrainConfig + +config = TrainConfig( + model_name="air_raid", + project_name="air_raid-example", + timestamp = 1_000_000, + num_envs = 4 +) +config.device = torch.device("cpu") +algo_config = AlgoConfig(ent_coef=0.0) + +trainer = easy_train_ppo("ALE/AirRaid-v5", config, algo_config) +trainer.train(use_wandb=True) +trainer.try_agent() diff --git a/examples/hopper_mujoco_immediate.py b/examples/hopper_mujoco_immediate.py index 803d03d..fc8543c 100644 --- a/examples/hopper_mujoco_immediate.py +++ b/examples/hopper_mujoco_immediate.py @@ -8,8 +8,7 @@ from __future__ import annotations import os -os.environ["MUJOCO_GL"] = "egl" -os.environ["PYOPENGL_PLATFORM"] = "egl" + import argparse @@ -350,7 +349,7 @@ def run_immediate( env = vectorize_env(Hopper2D, num_envs=cfg.num_envs) obs_dim, act_dim, obs_n, act_n, is_discrete = get_obs_act(env) agent = ActorCriticAgent(obs_n, act_n, is_discrete, hidden_layer).to(cfg.device) - normalizer = NormMeanStd((int(obs_n),), device=cfg.device) + normalizer = NormMeanStd(obs_dim, device=cfg.device) if resume_path is not None: ckpt = torch.load(resume_path, map_location=cfg.device, weights_only=False) @@ -491,4 +490,9 @@ def main() -> None: if __name__ == "__main__": + # Standalone runs may target headless EGL; set only here so importing + # this module (e.g. from tests) never poisons os.environ for spawned + # child processes that import mujoco fresh (CI child crash). + os.environ.setdefault("MUJOCO_GL", "egl") + os.environ.setdefault("PYOPENGL_PLATFORM", "egl") main() diff --git a/examples/pendulum.py b/examples/pendulum.py index 6069ae1..473d4de 100644 --- a/examples/pendulum.py +++ b/examples/pendulum.py @@ -6,8 +6,7 @@ model_name="Pendulum", project_name="Pendulum-example", timestamp = 1_000_000, - num_envs = 4, - profile = True + num_envs = 4 ) config.device = torch.device("cpu") algo_config = AlgoConfig(ent_coef=0.0) diff --git a/tests/env/test_hopper_immediate.py b/tests/env/test_hopper_immediate.py index ab03af1..46b5da6 100644 --- a/tests/env/test_hopper_immediate.py +++ b/tests/env/test_hopper_immediate.py @@ -73,7 +73,7 @@ def test_vectorize_class(self) -> None: obs, _ = env.reset(seed=seeds) assert obs.shape[0] == 2 obs_dim, act_dim, obs_n, act_n, is_discrete = get_obs_act(env) - assert obs_n == 11 + assert obs_n == 11 # vector obs -> flat int (matches the name) assert act_n == 3 assert is_discrete is False assert act_dim == (3,) diff --git a/tests/test_benchmark_scaling.py b/tests/test_benchmark_scaling.py index 19f605f..bd2c0dd 100644 --- a/tests/test_benchmark_scaling.py +++ b/tests/test_benchmark_scaling.py @@ -1199,8 +1199,8 @@ def test_torchrl_ppo_e2e_trunk_count(self, monkeypatch) -> None: policy_calls = [0] orig_wiring = pp.build_torchrl_ppo_wiring - def wiring_spy(trunk, cfg, updates): - wiring = orig_wiring(trunk, cfg, updates) + def wiring_spy(trunk, cfg, updates, **kwargs): + wiring = orig_wiring(trunk, cfg, updates, **kwargs) wiring["collector_policy"].register_forward_hook( lambda *_: policy_calls.__setitem__(0, policy_calls[0] + 1)) return wiring @@ -1208,8 +1208,8 @@ def wiring_spy(trunk, cfg, updates): monkeypatch.setattr(pp, "build_torchrl_ppo_wiring", wiring_spy) class CountingTrunk(ref.TorchRLSharedTrunk): - def __init__(self, sd=None): - super().__init__(sd) + def __init__(self, sd=None, **kwargs): + super().__init__(sd, **kwargs) self.trunk.register_forward_hook( lambda *_: counter.__setitem__(0, counter[0] + 1)) @@ -1698,28 +1698,28 @@ def boom(*_args, **_kwargs): class TestEnvAgnosticAdaptation: """Env-agnostic + continuous-action adaptation of the shared reference - policy (HumanoidStandup-v4 as the continuous case).""" + policy (HumanoidStandup-v5 as the continuous case).""" def test_env_slug_and_dims(self) -> None: - assert schema.env_slug("HumanoidStandup-v4") == "humanoidstandupv4" + assert schema.env_slug("HumanoidStandup-v5") == "humanoidstandupv5" 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,) + d = ref.env_dims("HumanoidStandup-v5") # gym.make; lru-cached + assert d.obs_shape == (348,) 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, + sd = ref.build_reference_state_dict(4242, obs_dim=348, act_dim=17, is_discrete=False) - assert sd["extract_layer.0.weight"].shape == (64, 376) + assert sd["extract_layer.0.weight"].shape == (64, 348) 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) + trunk = ref.TorchRLSharedTrunk(sd, obs_dim=348, act_dim=17) assert isinstance(trunk.log_std, torch.nn.Parameter) assert any(p is trunk.log_std for p in trunk.parameters()) @@ -1735,19 +1735,19 @@ def test_continuous_cross_framework_equivalence(self) -> None: from benchmarks.scalability.collection import _TorchRLCollectPolicy - sd = ref.build_reference_state_dict(4242, obs_dim=376, act_dim=17, + sd = ref.build_reference_state_dict(4242, obs_dim=348, act_dim=17, is_discrete=False) agent = ref.build_deterministic_zerorl_agent( - sd, obs_dim=376, act_dim=17, is_discrete=False) + sd, obs_dim=348, 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=(348,), 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) + trunk = ref.TorchRLSharedTrunk(sd, obs_dim=348, act_dim=17) torchrl_policy = _TorchRLCollectPolicy(trunk) torch.manual_seed(0) - obs = torch.randn(8, 376) + obs = torch.randn(8, 348) z = agent.get_action(obs) s_actions, s_values, s_lp, s_ent = sb3.forward_collection(obs) @@ -1777,8 +1777,8 @@ def test_env_flag_and_auto_out(self, monkeypatch, tmp_path) -> None: 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" + ["--env", "HumanoidStandup-v5"]) + assert parsed.env == "HumanoidStandup-v5" # resources report/plots expose build_parser: --env-slug default "" # + accepted override @@ -1812,7 +1812,7 @@ def stub_fn(*_args, __cap=captured, **kwargs): 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" + assert ("expA_" + schema.env_slug("HumanoidStandup-v5") + + ".jsonl") == "expA_humanoidstandupv5.jsonl" + assert ("resources_" + schema.env_slug("HumanoidStandup-v5") + + ".jsonl") == "resources_humanoidstandupv5.jsonl" diff --git a/tests/test_encoder.py b/tests/test_encoder.py new file mode 100644 index 0000000..c2908fb --- /dev/null +++ b/tests/test_encoder.py @@ -0,0 +1,95 @@ +"""Unit tests for zerorl.helpers.encoder.ObservationEncoder.""" + +import pytest +import torch +from torch import nn + +from zerorl.helpers.encoder import ObservationEncoder + + +class TestVectorPaths: + def test_int_and_tuple_shapes_equivalent(self) -> None: + e_int = ObservationEncoder(8) + e_tup = ObservationEncoder((8,)) + x = torch.randn(5, 8) + assert e_int.output_dim == e_tup.output_dim == 8 + assert torch.equal(e_int(x), e_tup(x)) + # vector encoders are pure flattening: ZERO parameters, which is + # what keeps the shared-trunk reference/sd contract intact + assert sum(p.numel() for p in e_int.parameters()) == 0 + assert e_int.is_image is False and e_int.image_layout is None + + def test_multidim_vector_flattens(self) -> None: + enc = ObservationEncoder((4, 5)) + assert enc.output_dim == 20 + out = enc(torch.randn(2, 4, 5)) + assert out.shape == (2, 20) + + def test_single_sample_gets_batched(self) -> None: + enc = ObservationEncoder(8) + assert enc(torch.randn(8)).shape == (1, 8) + assert enc(torch.randn(3, 8)).shape == (3, 8) + + def test_invalid_shapes_and_scale_raise(self) -> None: + with pytest.raises(ValueError, match="Invalid observation shape"): + ObservationEncoder(0) + with pytest.raises(ValueError, match="Invalid observation shape"): + ObservationEncoder((-3,)) + with pytest.raises(ValueError, match="image_scale"): + ObservationEncoder(8, image_scale=0) + + +class TestImagePaths: + def test_hwc_and_chw_equivalent_after_layout(self) -> None: + torch.manual_seed(0) + chw = ObservationEncoder((3, 84, 84), image_layout="CHW") + hwc = ObservationEncoder((84, 84, 3), image_layout="HWC") + hwc.features.load_state_dict(chw.features.state_dict()) + img = torch.rand(2, 3, 84, 84) * 255.0 + out_chw = chw(img) + out_hwc = hwc(img.permute(0, 2, 3, 1).contiguous()) + assert out_chw.shape == out_hwc.shape + assert torch.allclose(out_chw, out_hwc, atol=1e-6) + + def test_auto_layout_detection(self) -> None: + assert ObservationEncoder((3, 84, 84)).image_layout == "CHW" + assert ObservationEncoder((84, 84, 3)).image_layout == "HWC" + # ambiguous: both first and last look like channels + with pytest.raises(ValueError, match="Cannot infer image layout"): + ObservationEncoder((3, 8, 3)) + + def test_bad_explicit_layout_raises(self) -> None: + with pytest.raises(ValueError, match="image_layout"): + ObservationEncoder((3, 84, 84), image_layout="XYZ") + + def test_pixel_scale_applied_only_when_enabled(self) -> None: + class Recorder(nn.Module): + def __init__(self) -> None: + super().__init__() + self.seen = None + + def forward(self, x): + self.seen = x + return x.flatten(1) + + enc = ObservationEncoder((3, 84, 84), image_scale=255.0) + rec = Recorder() + enc.features = rec + enc(torch.full((1, 3, 84, 84), 255.0)) + assert rec.seen is not None and float(rec.seen.max()) == pytest.approx(1.0) + + enc_ns = ObservationEncoder((3, 84, 84), image_scale=None) + rec2 = Recorder() + enc_ns.features = rec2 + enc_ns(torch.full((1, 3, 84, 84), 255.0)) + assert float(rec2.seen.max()) == pytest.approx(255.0) + + def test_too_small_image_raises(self) -> None: + with pytest.raises(ValueError, match="too small"): + ObservationEncoder((1, 4, 4), image_layout="CHW") + + def test_output_dim_matches_forward(self) -> None: + enc = ObservationEncoder((1, 84, 84)) + out = enc(torch.zeros(1, 1, 84, 84)) + assert enc.output_dim == out.shape[-1] + assert enc.n_features == enc.output_dim diff --git a/tests/test_factory.py b/tests/test_factory.py index 994900e..a7b732f 100644 --- a/tests/test_factory.py +++ b/tests/test_factory.py @@ -2,6 +2,7 @@ import pytest import torch +from torch import nn from zerorl.helpers.factory import ( get_env, get_actor_critic_buffer, ActorCriticAgent, get_policy_buffer, get_replay_buffer, PolicyAgent, @@ -196,4 +197,63 @@ def test_device_property(self, device) -> None: cfg.num_envs = 1 cfg.device = device buf = Buffer(capacity=cfg.rollout_steps, num_envs=cfg.num_envs, schema={"state": (4,)}, device=cfg.device) - assert buf.device.type == device.type \ No newline at end of file + assert buf.device.type == device.type + + +class TestEncoderIntegration: + """Factory agents run ObservationEncoder in front of the trunk + (commits 699d28d/17286bf/2983ae5); pin the contracts that keep the + cross-framework reference weights valid.""" + + def test_vector_agent_encoder_wiring_and_param_count(self) -> None: + from zerorl.helpers.factory import ActorCriticAgent + + agent = ActorCriticAgent(8, 4, is_discrete=True, hidden_dim=64) + assert agent.encoder.output_dim == agent.encoder_dim == 8 + assert sum(p.numel() for p in agent.encoder.parameters()) == 0 + # encoder is invoked exactly once per forward + calls = [0] + agent.encoder.register_forward_hook( + lambda *_: calls.__setitem__(0, calls[0] + 1)) + state = torch.randn(5, 8) + with torch.no_grad(): + logits, value = agent(state) + assert calls[0] == 1 + assert logits.shape == (5, 4) and value.shape == (5, 1) + # forward == manual chain over the same modules + with torch.no_grad(): + x = agent.extract_layer(agent.encoder(state)) + m_logits, m_value = agent.actor(x), agent.critic(x) + assert torch.equal(logits, m_logits) and torch.equal(value, m_value) + # vector agents add NO parameters -> the 5061 shared-trunk contract + # (4736 trunk + 260 actor + 65 critic) that benchmarks rely on holds + assert sum(p.numel() for p in agent.parameters()) == 5061 + + def test_image_agent_uses_naturecnn_and_ortho_init(self) -> None: + from zerorl.helpers.factory import ActorCriticAgent + + agent = ActorCriticAgent((1, 84, 84), 4, is_discrete=True, + hidden_dim=64) + convs = [m for m in agent.encoder.features + if isinstance(m, nn.Conv2d)] + assert len(convs) == 3 + for conv in convs: + assert torch.isfinite(conv.weight).all() + assert conv.weight.abs().sum() > 0 # orthogonally inited + assert conv.bias is not None and float(conv.bias.abs().sum()) == 0 + # encoder output feeds the trunk + assert agent.encoder.output_dim == agent.extract_layer[0].in_features + with torch.no_grad(): + logits, value = agent(torch.rand(2, 1, 84, 84)) + assert logits.shape == (2, 4) and value.shape == (2, 1) + + def test_policy_agent_encoder_smoke(self) -> None: + from zerorl.helpers.factory import PolicyAgent + + agent = PolicyAgent(8, 4, is_discrete=True, hidden_dim=64) + assert agent.encoder.output_dim == 8 + with torch.no_grad(): + out = agent(torch.randn(3, 8)) + # PolicyAgent.forward returns logits only + assert torch.is_tensor(out) and out.shape == (3, 4) + diff --git a/tests/test_function.py b/tests/test_function.py index d3d52ae..1288986 100644 --- a/tests/test_function.py +++ b/tests/test_function.py @@ -130,6 +130,32 @@ def test_discrete_env_action_shape_is_empty(self, cartpole_env) -> None: _, act_shape, _, _, _ = get_obs_act(cartpole_env) assert act_shape == () + def test_box_env_obs_n_is_flat_int(self, pendulum_env) -> None: + """Contract: for vector obs obs_n is the FLAT size (int) — matches + the name; obs_shape keeps the tuple.""" + obs_shape, act_shape, obs_n, act_n, is_discrete = get_obs_act( + pendulum_env) + assert obs_shape == (3,) + assert isinstance(obs_n, int) and obs_n == 3 + assert is_discrete is False and act_n == 1 + + def test_image_env_obs_n_is_rank3_tuple(self) -> None: + """Rank-3 image obs: obs_n is the (C, H, W) tuple for CNNs.""" + import types + + import gymnasium as gym2 + import numpy as np2 + + fake = types.SimpleNamespace( + observation_space=gym2.spaces.Box(0, 255, (84, 84, 3), + dtype=np2.uint8), + action_space=gym2.spaces.Discrete(4), + ) + obs_shape, _, obs_n, act_n, is_discrete = get_obs_act(fake) + assert obs_shape == (84, 84, 3) + assert obs_n == (84, 84, 3) and isinstance(obs_n, tuple) + assert is_discrete is True and act_n == 4 + def test_discrete_env_obs_dim(self, cartpole_env) -> None: obs_shape, _, obs_n, _, _ = get_obs_act(cartpole_env) assert obs_shape == (4,) diff --git a/tests/test_resources.py b/tests/test_resources.py index 6d5218f..7e66715 100644 --- a/tests/test_resources.py +++ b/tests/test_resources.py @@ -674,8 +674,8 @@ def test_resources_env_flag_and_slug() -> None: args = build_parser().parse_args([]) assert args.env == "LunarLander-v3" - args = build_parser().parse_args(["--env", "HumanoidStandup-v4"]) - assert args.env == "HumanoidStandup-v4" + args = build_parser().parse_args(["--env", "HumanoidStandup-v5"]) + assert args.env == "HumanoidStandup-v5" # 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" + assert f"{env_slug('HumanoidStandup-v5')}_collect_zeroRL_rep0.jsonl" == \ + "humanoidstandupv5_collect_zeroRL_rep0.jsonl" diff --git a/tests/test_sweeps.py b/tests/test_sweeps.py new file mode 100644 index 0000000..12ebc47 --- /dev/null +++ b/tests/test_sweeps.py @@ -0,0 +1,69 @@ +"""Unit tests for zerorl.helpers.sweeps (wandb/optuna wrappers).""" + +import sys +import types + +import pytest + +from zerorl.helpers.sweeps import run_optuna_sweep, run_wandb_sweep + + +def test_wandb_missing_raises_install_hint(monkeypatch) -> None: + monkeypatch.setitem(sys.modules, "wandb", None) + with pytest.raises(ImportError, match="wandb"): + run_wandb_sweep({"method": "grid"}, lambda: None, "proj") + + +def test_optuna_missing_raises_install_hint(monkeypatch) -> None: + monkeypatch.setitem(sys.modules, "optuna", None) + with pytest.raises(ImportError, match="optuna"): + run_optuna_sweep(lambda trial: 0.0) + + +def test_wandb_delegates_to_sweep_and_agent(monkeypatch) -> None: + calls = {} + mod = types.ModuleType("wandb") + + def sweep(**kwargs): + calls["sweep"] = kwargs + return "sweep-id-1" + + def agent(sweep_id, function=None, count=None): + calls["agent"] = (sweep_id, count, function) + + mod.sweep = sweep + mod.agent = agent + monkeypatch.setitem(sys.modules, "wandb", mod) + + fn = lambda: None # noqa: E731 + run_wandb_sweep({"method": "grid"}, fn, "my-proj", num_trials=3) + assert calls["sweep"] == {"sweep": {"method": "grid"}, "project": "my-proj"} + assert calls["agent"] == ("sweep-id-1", 3, fn) + + +def test_optuna_delegates_to_create_study_and_optimize(monkeypatch) -> None: + calls = {} + mod = types.ModuleType("optuna") + + class _Study: + best_params = {"lr": 0.01} + + # NOTE: class-attribute callables receive `self` as first arg — + # declaring it is required or the bound self lands in `fn`. + def optimize(self, fn, n_trials=None, **kwargs): + calls["optimize"] = (fn, n_trials) + + def create_study(**kwargs): + calls["create_study"] = kwargs + return _Study() + + mod.create_study = create_study + monkeypatch.setitem(sys.modules, "optuna", mod) + + def objective(trial): + return 1.0 + + best = run_optuna_sweep(objective, direction="minimize", num_trials=5) + assert calls["create_study"] == {"direction": "minimize"} + assert calls["optimize"] == (objective, 5) + assert best == {"lr": 0.01} diff --git a/zerorl/functions.py b/zerorl/functions.py index 5c3c67b..442225f 100644 --- a/zerorl/functions.py +++ b/zerorl/functions.py @@ -1,6 +1,6 @@ """Utility functions for RL training. -Provides vectorize_env() for environment wrapping, env_step() forfunc +Provides vectorize_env() for environment wrapping, env_step() for agent-env interaction, and helpers for state processing, seeding, and action conversion. """ @@ -15,6 +15,7 @@ import imageio from typing import Any, Callable from gymnasium import spaces +from gymnasium.wrappers import AtariPreprocessing, FrameStackObservation from gymnasium.vector import AutoresetMode, SyncVectorEnv from torch import Tensor from torch.nn import Parameter @@ -24,30 +25,95 @@ from zerorl.config import TrainConfig -def vectorize_env(env_spec: str | Callable | BaseEnv, *, num_envs: int = 1, render_mode: str | None = None) -> SyncVectorEnv: +ATARI_PREPROCESS: dict[str, Any] = dict( + noop_max = 30, + frame_skip = 4, + screen_size = 84, + terminal_on_life_loss = False, + grayscale_obs = True, + grayscale_newaxis = False, + scale_obs = False) + +def vectorize_env(env_spec: str | Callable | BaseEnv, + *, + num_envs: int = 1, + atari_preprocess: dict | None = None, + frame_stack: int | None = None, + extra_wrappers: list[tuple[type, dict]] | None = None, + env_kwargs: dict | None = None, + render_mode: str | None = None, + autoreset_mode: AutoresetMode = AutoresetMode.SAME_STEP + ) -> SyncVectorEnv: """Wrap an env spec into a SyncVectorEnv with SAME_STEP autoreset. Args: env_spec: Gymnasium env ID string, BaseEnv class/instance, or callable. num_envs: Number of parallel environments. render_mode: Render mode for the environment. + atari_preprocess: AtariPreprocessing kwargs for ALE/ envs. + frame_stack: FrameStackObservation stack size. None (default) = + AUTO: stack 4 frames only for ALE/ Atari envs (their standard + protocol) and leave every other env's observations untouched; + pass an int to opt in for non-Atari envs explicitly. Returns: SyncVectorEnv wrapping num_envs independent copies. """ + is_atari = False + atari_kwargs = dict(ATARI_PREPROCESS) + atari_kwargs.update(atari_preprocess or {}) + extra_wrappers = list(extra_wrappers or []) + env_kwargs = dict(env_kwargs or {}) + + def _ensure_ale_registered(): + try: + import ale_py + gym.register_envs(ale_py) + except ImportError: + pass + def make_env_fn() -> Callable: def _init(): if isinstance(env_spec, str): - env = gym.make(env_spec, render_mode = render_mode) + if env_spec.startswith("ALE/"): + _ensure_ale_registered() + nonlocal is_atari + env_kwargs.setdefault("frameskip", 1) + is_atari = True + env = gym.make(env_spec, render_mode = render_mode, **env_kwargs) elif isinstance(env_spec, type): env = env_spec() elif callable(env_spec): env = env_spec() else: env = copy.deepcopy(env_spec) + + if is_atari: + env = AtariPreprocessing(env, **atari_kwargs) + # stacking: explicit int = opt-in; None = auto (Atari only) so + # non-Atari observation shapes stay untouched by default + stack = frame_stack if frame_stack is not None else (4 if is_atari else None) + if stack is not None and stack > 0: + env = FrameStackObservation(env, stack_size=stack) + for wrapper_cls, wrapper_kwargs in extra_wrappers: + env = wrapper_cls(env, **wrapper_kwargs) return env return _init - return gym.vector.SyncVectorEnv([make_env_fn() for _ in range(num_envs)], autoreset_mode=AutoresetMode.SAME_STEP) + + venv = gym.vector.SyncVectorEnv([make_env_fn() for _ in range(num_envs)], autoreset_mode=autoreset_mode) + effective_config = { + "env_spec": str(env_spec), + "num_envs": num_envs, + "render_mode": render_mode, + "env_kwargs": dict(env_kwargs), + "atari_preprocess": dict(atari_kwargs) if is_atari else None, + "frame_stack": frame_stack if frame_stack is not None else (4 if is_atari else None), + "extra_wrappers": [(w.__name__, dict(k)) for w, k in extra_wrappers], + "autoreset_mode": autoreset_mode.name, + } + venv.effective_config = lambda: dict(effective_config) + return venv + def env_step(env: Any, agent: BaseAgent, state: Tensor) -> dict[str, Any]: """Run one agent-environment step: get_action -> env.step -> return transition dict.""" @@ -294,7 +360,10 @@ def get_obs_act(env: SyncVectorEnv) -> Any: Returns: 5-tuple of (obs_shape, act_shape, obs_n, act_n, is_discrete). obs_shape/act_shape are raw shape tuples from the spaces. - obs_n is the observation dim (or the Image space for image obs). + obs_n matches its name for vector obs: the FLAT observation size as + an int (np.prod of the shape — feed straight to Linear/NormMeanStd). + For rank-3 image obs obs_n is the (C, H, W) TUPLE instead, because + CNNs need the rank-3 shape, not a flat number. act_n is the action dim (int for discrete, product-of-shapes for continuous). is_discrete is a bool. """ @@ -312,8 +381,10 @@ def get_obs_act(env: SyncVectorEnv) -> Any: else: act_n = int(np.prod(act_dim.shape)) #type: ignore - obs_n = obs_dim.shape - + if len(obs_dim.shape) > 2: # type: ignore + obs_n = obs_dim.shape # image obs: (C, H, W) tuple — CNN needs the rank-3 shape + else: + obs_n = int(np.prod(obs_dim.shape)) # type: ignore # flat size (matches the name) return (obs_dim.shape, act_dim.shape, obs_n, act_n, is_discrete) def actions_limits(env: SyncVectorEnv, agent: BaseAgent) -> tuple[Any, Any] | None: