Upload folder using huggingface_hub

.gitattributes

@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+replay.mp4 filter=lfs diff=lfs merge=lfs -text

.summary/0/events.out.tfevents.1760762395.7f51b3d89358

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1a612b9161c98815d98ec295cb5242ad6e1748adc1508a9d2d24f8c46085dbc4
+size 357576

.summary/0/events.out.tfevents.1760812661.9740909df8d5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:96689dd679a964ebf1d2d3ae5cea47629e0cec47deafa7e67e4c234db1b0ccc1
+size 4168

.summary/0/events.out.tfevents.1760832021.6d115e46a7aa

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:29ed3e1415d9127daa093a8b606013236a0d6d4ecae71711352eacd74a34623f
+size 4132

.summary/0/events.out.tfevents.1760834099.c4fccf2fafc1

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7cf930fc0dc32cf6cebeebc32fad78f0538ce0975f088543a4758d9349136075
+size 1308950

README.md

@@ -0,0 +1,56 @@
+---
+library_name: sample-factory
+tags:
+- deep-reinforcement-learning
+- reinforcement-learning
+- sample-factory
+model-index:
+- name: APPO
+  results:
+  - task:
+      type: reinforcement-learning
+      name: reinforcement-learning
+    dataset:
+      name: doom_deadly_corridor
+      type: doom_deadly_corridor
+    metrics:
+    - type: mean_reward
+      value: 8.92 +/- 7.43
+      name: mean_reward
+      verified: false
+---
+
+A(n) **APPO** model trained on the **doom_deadly_corridor** environment.
+
+This model was trained using Sample-Factory 2.0: https://github.com/alex-petrenko/sample-factory.
+Documentation for how to use Sample-Factory can be found at https://www.samplefactory.dev/
+
+
+## Downloading the model
+
+After installing Sample-Factory, download the model with:
+```
+python -m sample_factory.huggingface.load_from_hub -r elliemci/deadly_corridor_experiment
+```
+
+    
+## Using the model
+
+To run the model after download, use the `enjoy` script corresponding to this environment:
+```
+python -m <path.to.enjoy.module> --algo=APPO --env=doom_deadly_corridor --train_dir=./train_dir --experiment=deadly_corridor_experiment
+```
+
+
+You can also upload models to the Hugging Face Hub using the same script with the `--push_to_hub` flag.
+See https://www.samplefactory.dev/10-huggingface/huggingface/ for more details
+    
+## Training with this model
+
+To continue training with this model, use the `train` script corresponding to this environment:
+```
+python -m <path.to.train.module> --algo=APPO --env=doom_deadly_corridor --train_dir=./train_dir --experiment=deadly_corridor_experiment --restart_behavior=resume --train_for_env_steps=10000000000
+```
+
+Note, you may have to adjust `--train_for_env_steps` to a suitably high number as the experiment will resume at the number of steps it concluded at.
+    

checkpoint_p0/best_000001047_4288512_reward_15.226.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:85e01ad2e691bb821d8f1613b919759d5fb2bbd4eabbc66be14ec12a1c25d584
+size 34966380

checkpoint_p0/checkpoint_000000973_3985408.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:50c8e46872f33efcf183be9bf2694f977e98b3e1115dd74e53805ab47eefab52
+size 34966818

checkpoint_p0/checkpoint_000001030_4218880.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d260347365353c92ee867c53b0b215d2521e42b4b81cc5140c41981d766fc76c
+size 34966818

config.json

@@ -0,0 +1,149 @@
+{
+  "help": false,
+  "algo": "APPO",
+  "env": "doom_deadly_corridor",
+  "experiment": "deadly_corridor_experiment",
+  "train_dir": "train_dir",
+  "restart_behavior": "resume",
+  "device": "gpu",
+  "seed": null,
+  "num_policies": 1,
+  "async_rl": true,
+  "serial_mode": false,
+  "batched_sampling": false,
+  "num_batches_to_accumulate": 2,
+  "worker_num_splits": 2,
+  "policy_workers_per_policy": 1,
+  "max_policy_lag": 1000,
+  "num_workers": 1,
+  "num_envs_per_worker": 4,
+  "batch_size": 1024,
+  "num_batches_per_epoch": 1,
+  "num_epochs": 1,
+  "rollout": 32,
+  "recurrence": 32,
+  "shuffle_minibatches": false,
+  "gamma": 0.99,
+  "reward_scale": 1.0,
+  "reward_clip": 1000.0,
+  "value_bootstrap": false,
+  "normalize_returns": true,
+  "exploration_loss_coeff": 0.001,
+  "value_loss_coeff": 0.5,
+  "kl_loss_coeff": 0.0,
+  "exploration_loss": "symmetric_kl",
+  "gae_lambda": 0.95,
+  "ppo_clip_ratio": 0.1,
+  "ppo_clip_value": 0.2,
+  "with_vtrace": false,
+  "vtrace_rho": 1.0,
+  "vtrace_c": 1.0,
+  "optimizer": "adam",
+  "adam_eps": 1e-06,
+  "adam_beta1": 0.9,
+  "adam_beta2": 0.999,
+  "max_grad_norm": 4.0,
+  "learning_rate": 0.0001,
+  "lr_schedule": "constant",
+  "lr_schedule_kl_threshold": 0.008,
+  "lr_adaptive_min": 1e-06,
+  "lr_adaptive_max": 0.01,
+  "obs_subtract_mean": 0.0,
+  "obs_scale": 255.0,
+  "normalize_input": true,
+  "normalize_input_keys": null,
+  "decorrelate_experience_max_seconds": 0,
+  "decorrelate_envs_on_one_worker": true,
+  "actor_worker_gpus": [],
+  "set_workers_cpu_affinity": true,
+  "force_envs_single_thread": false,
+  "default_niceness": 0,
+  "log_to_file": true,
+  "experiment_summaries_interval": 60,
+  "flush_summaries_interval": 10,
+  "stats_avg": 10,
+  "summaries_use_frameskip": true,
+  "heartbeat_interval": 20,
+  "heartbeat_reporting_interval": 600,
+  "train_for_env_steps": 5000000,
+  "train_for_seconds": 10000000000,
+  "save_every_sec": 120,
+  "keep_checkpoints": 2,
+  "load_checkpoint_kind": "latest",
+  "save_milestones_sec": -1,
+  "save_best_every_sec": 5,
+  "save_best_metric": "reward",
+  "save_best_after": 100000,
+  "benchmark": false,
+  "encoder_mlp_layers": [
+    512,
+    512
+  ],
+  "encoder_conv_architecture": "convnet_simple",
+  "encoder_conv_mlp_layers": [
+    512
+  ],
+  "use_rnn": true,
+  "rnn_size": 512,
+  "rnn_type": "gru",
+  "rnn_num_layers": 1,
+  "decoder_mlp_layers": [],
+  "nonlinearity": "elu",
+  "policy_initialization": "orthogonal",
+  "policy_init_gain": 1.0,
+  "actor_critic_share_weights": true,
+  "adaptive_stddev": true,
+  "continuous_tanh_scale": 0.0,
+  "initial_stddev": 1.0,
+  "use_env_info_cache": false,
+  "env_gpu_actions": false,
+  "env_gpu_observations": true,
+  "env_frameskip": 4,
+  "env_framestack": 1,
+  "pixel_format": "CHW",
+  "use_record_episode_statistics": false,
+  "with_wandb": true,
+  "wandb_user": null,
+  "wandb_project": "sample_factory",
+  "wandb_group": null,
+  "wandb_job_type": "SF",
+  "wandb_tags": [],
+  "with_pbt": false,
+  "pbt_mix_policies_in_one_env": true,
+  "pbt_period_env_steps": 5000000,
+  "pbt_start_mutation": 20000000,
+  "pbt_replace_fraction": 0.3,
+  "pbt_mutation_rate": 0.15,
+  "pbt_replace_reward_gap": 0.1,
+  "pbt_replace_reward_gap_absolute": 1e-06,
+  "pbt_optimize_gamma": false,
+  "pbt_target_objective": "true_objective",
+  "pbt_perturb_min": 1.1,
+  "pbt_perturb_max": 1.5,
+  "num_agents": -1,
+  "num_humans": 0,
+  "num_bots": -1,
+  "start_bot_difficulty": null,
+  "timelimit": null,
+  "res_w": 128,
+  "res_h": 72,
+  "wide_aspect_ratio": false,
+  "eval_env_frameskip": 1,
+  "fps": 35,
+  "command_line": "--env=doom_deadly_corridor --num_workers=8 --num_envs_per_worker=4 --train_for_env_steps=5000000 --experiment=deadly_corridor_experiment --train_dir=train_dir --with_wandb=True --experiment_summaries_interval=60 --flush_summaries_interval=10 --stats_avg=10",
+  "cli_args": {
+    "env": "doom_deadly_corridor",
+    "experiment": "deadly_corridor_experiment",
+    "train_dir": "train_dir",
+    "num_workers": 8,
+    "num_envs_per_worker": 4,
+    "experiment_summaries_interval": 60,
+    "flush_summaries_interval": 10,
+    "stats_avg": 10,
+    "train_for_env_steps": 5000000,
+    "with_wandb": true
+  },
+  "git_hash": "unknown",
+  "git_repo_name": "not a git repository",
+  "wandb_unique_id": "deadly_corridor_experiment_20251018_043945_196881"
+}

log.txt

@@ -0,0 +1,56 @@
+Gym has been unmaintained since 2022 and does not support NumPy 2.0 amongst other critical functionality.
+Please upgrade to Gymnasium, the maintained drop-in replacement of Gym, or contact the authors of your software and request that they upgrade.
+See the migration guide at https://gymnasium.farama.org/introduction/migration_guide/ for additional information.
+[2025-10-19 00:31:54,997][05777] register_encoder_factory: <function make_vizdoom_encoder at 0x785a3660e3e0>
+[2025-10-19 00:31:55,569][05777] Loading existing experiment configuration from train_dir/deadly_corridor_experiment/config.json
+[2025-10-19 00:31:55,570][05777] Overriding arg 'train_for_env_steps' with value 5000000 passed from command line
+[2025-10-19 00:31:55,578][05777] Experiment dir train_dir/deadly_corridor_experiment already exists!
+[2025-10-19 00:31:55,578][05777] Resuming existing experiment from train_dir/deadly_corridor_experiment...
+[2025-10-19 00:31:55,578][05777] Weights and Biases integration enabled. Project: sample_factory, user: None, group: None, unique_id: deadly_corridor_experiment_20251018_043945_196881
+[2025-10-19 00:31:58,148][05777] Initializing WandB...
+wandb: WARNING `start_method` is deprecated and will be removed in a future version of wandb. This setting is currently non-functional and safely ignored.
+[2025-10-19 00:31:58,172][05777] Exception thrown when attempting to run <function init_wandb.<locals>.init_wandb_func at 0x785a36525d00>, attempt 0 out of 3
+[2025-10-19 00:31:59,182][05777] Exception thrown when attempting to run <function init_wandb.<locals>.init_wandb_func at 0x785a36525d00>, attempt 1 out of 3
+[2025-10-19 00:32:01,192][05777] Exception thrown when attempting to run <function init_wandb.<locals>.init_wandb_func at 0x785a36525d00>, attempt 2 out of 3
+[2025-10-19 00:32:05,202][05777] Could not initialize WandB! api_key not configured (no-tty). call wandb.login(key=[your_api_key])
+Traceback (most recent call last):
+  File "<frozen runpy>", line 198, in _run_module_as_main
+  File "<frozen runpy>", line 88, in _run_code
+  File "/usr/local/lib/python3.12/dist-packages/sf_examples/vizdoom/train_vizdoom.py", line 48, in <module>
+    sys.exit(main())
+             ^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sf_examples/vizdoom/train_vizdoom.py", line 43, in main
+    status = run_rl(cfg)
+             ^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/train.py", line 32, in run_rl
+    cfg, runner = make_runner(cfg)
+                  ^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/train.py", line 23, in make_runner
+    runner = runner_cls(cfg)
+             ^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/runners/runner_parallel.py", line 17, in __init__
+    super().__init__(cfg)
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/runners/runner.py", line 135, in __init__
+    init_wandb(self.cfg)  # should be done before writers are initialized
+    ^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/utils/wandb_utils.py", line 51, in init_wandb
+    init_wandb_func()
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/utils/utils.py", line 181, in newfn
+    return func(*args, **kwargs)
+           ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/utils/wandb_utils.py", line 36, in init_wandb_func
+    wandb.init(
+  File "/usr/local/lib/python3.12/dist-packages/wandb/sdk/wandb_init.py", line 1601, in init
+    wandb._sentry.reraise(e)
+  File "/usr/local/lib/python3.12/dist-packages/wandb/analytics/sentry.py", line 162, in reraise
+    raise exc.with_traceback(sys.exc_info()[2])
+  File "/usr/local/lib/python3.12/dist-packages/wandb/sdk/wandb_init.py", line 1523, in init
+    wi.maybe_login(init_settings)
+  File "/usr/local/lib/python3.12/dist-packages/wandb/sdk/wandb_init.py", line 190, in maybe_login
+    wandb_login._login(
+  File "/usr/local/lib/python3.12/dist-packages/wandb/sdk/wandb_login.py", line 320, in _login
+    key, key_status = wlogin.prompt_api_key(referrer=referrer)
+                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/wandb/sdk/wandb_login.py", line 245, in prompt_api_key
+    raise UsageError("api_key not configured (no-tty). call " + directive)
+wandb.errors.errors.UsageError: api_key not configured (no-tty). call wandb.login(key=[your_api_key])

replay.mp4

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:78d0dbdbd7d3639767d9304c62c8bb6b7fb30671e0ba60882ab0761fb1e42793
+size 1833187

sf_log.txt

@@ -0,0 +1,860 @@
+[2025-10-18 04:40:01,685][02528] Saving configuration to train_dir/deadly_corridor_experiment/config.json...
+[2025-10-18 04:40:01,697][02528] Rollout worker 0 uses device cpu
+[2025-10-18 04:40:01,699][02528] Rollout worker 1 uses device cpu
+[2025-10-18 04:40:01,701][02528] Rollout worker 2 uses device cpu
+[2025-10-18 04:40:01,703][02528] Rollout worker 3 uses device cpu
+[2025-10-18 04:40:01,705][02528] Rollout worker 4 uses device cpu
+[2025-10-18 04:40:01,707][02528] Rollout worker 5 uses device cpu
+[2025-10-18 04:40:01,709][02528] Rollout worker 6 uses device cpu
+[2025-10-18 04:40:01,710][02528] Rollout worker 7 uses device cpu
+[2025-10-18 04:40:01,854][02528] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-18 04:40:01,856][02528] InferenceWorker_p0-w0: min num requests: 2
+[2025-10-18 04:40:01,890][02528] Starting all processes...
+[2025-10-18 04:40:01,892][02528] Starting process learner_proc0
+[2025-10-18 04:40:01,971][02528] Starting all processes...
+[2025-10-18 04:40:01,982][02528] Starting process inference_proc0-0
+[2025-10-18 04:40:01,983][02528] Starting process rollout_proc0
+[2025-10-18 04:40:01,983][02528] Starting process rollout_proc1
+[2025-10-18 04:40:01,986][02528] Starting process rollout_proc2
+[2025-10-18 04:40:01,986][02528] Starting process rollout_proc3
+[2025-10-18 04:40:01,987][02528] Starting process rollout_proc4
+[2025-10-18 04:40:01,987][02528] Starting process rollout_proc5
+[2025-10-18 04:40:01,987][02528] Starting process rollout_proc6
+[2025-10-18 04:40:01,987][02528] Starting process rollout_proc7
+[2025-10-18 04:40:17,938][04351] Worker 0 uses CPU cores [0]
+[2025-10-18 04:40:18,215][04349] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-18 04:40:18,227][04349] Set environment var CUDA_VISIBLE_DEVICES to '0' (GPU indices [0]) for inference process 0
+[2025-10-18 04:40:18,312][04349] Num visible devices: 1
+[2025-10-18 04:40:18,325][04336] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-18 04:40:18,342][04336] Set environment var CUDA_VISIBLE_DEVICES to '0' (GPU indices [0]) for learning process 0
+[2025-10-18 04:40:18,389][04336] Num visible devices: 1
+[2025-10-18 04:40:18,393][04336] Starting seed is not provided
+[2025-10-18 04:40:18,396][04336] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-18 04:40:18,397][04336] Initializing actor-critic model on device cuda:0
+[2025-10-18 04:40:18,398][04336] RunningMeanStd input shape: (3, 72, 128)
+[2025-10-18 04:40:18,404][04336] RunningMeanStd input shape: (1,)
+[2025-10-18 04:40:18,421][04350] Worker 1 uses CPU cores [1]
+[2025-10-18 04:40:18,451][04336] ConvEncoder: input_channels=3
+[2025-10-18 04:40:18,453][04355] Worker 5 uses CPU cores [1]
+[2025-10-18 04:40:18,471][04352] Worker 2 uses CPU cores [0]
+[2025-10-18 04:40:18,623][04354] Worker 3 uses CPU cores [1]
+[2025-10-18 04:40:18,645][04357] Worker 7 uses CPU cores [1]
+[2025-10-18 04:40:18,647][04356] Worker 6 uses CPU cores [0]
+[2025-10-18 04:40:18,685][04353] Worker 4 uses CPU cores [0]
+[2025-10-18 04:40:18,772][04336] Conv encoder output size: 512
+[2025-10-18 04:40:18,773][04336] Policy head output size: 512
+[2025-10-18 04:40:18,822][04336] Created Actor Critic model with architecture:
+[2025-10-18 04:40:18,822][04336] ActorCriticSharedWeights(
+  (obs_normalizer): ObservationNormalizer(
+    (running_mean_std): RunningMeanStdDictInPlace(
+      (running_mean_std): ModuleDict(
+        (obs): RunningMeanStdInPlace()
+      )
+    )
+  )
+  (returns_normalizer): RecursiveScriptModule(original_name=RunningMeanStdInPlace)
+  (encoder): VizdoomEncoder(
+    (basic_encoder): ConvEncoder(
+      (enc): RecursiveScriptModule(
+        original_name=ConvEncoderImpl
+        (conv_head): RecursiveScriptModule(
+          original_name=Sequential
+          (0): RecursiveScriptModule(original_name=Conv2d)
+          (1): RecursiveScriptModule(original_name=ELU)
+          (2): RecursiveScriptModule(original_name=Conv2d)
+          (3): RecursiveScriptModule(original_name=ELU)
+          (4): RecursiveScriptModule(original_name=Conv2d)
+          (5): RecursiveScriptModule(original_name=ELU)
+        )
+        (mlp_layers): RecursiveScriptModule(
+          original_name=Sequential
+          (0): RecursiveScriptModule(original_name=Linear)
+          (1): RecursiveScriptModule(original_name=ELU)
+        )
+      )
+    )
+  )
+  (core): ModelCoreRNN(
+    (core): GRU(512, 512)
+  )
+  (decoder): MlpDecoder(
+    (mlp): Identity()
+  )
+  (critic_linear): Linear(in_features=512, out_features=1, bias=True)
+  (action_parameterization): ActionParameterizationDefault(
+    (distribution_linear): Linear(in_features=512, out_features=11, bias=True)
+  )
+)
+[2025-10-18 04:40:19,109][04336] Using optimizer <class 'torch.optim.adam.Adam'>
+[2025-10-18 04:40:21,851][02528] Heartbeat connected on Batcher_0
+[2025-10-18 04:40:21,856][02528] Heartbeat connected on InferenceWorker_p0-w0
+[2025-10-18 04:40:21,867][02528] Heartbeat connected on RolloutWorker_w0
+[2025-10-18 04:40:21,873][02528] Heartbeat connected on RolloutWorker_w1
+[2025-10-18 04:40:21,881][02528] Heartbeat connected on RolloutWorker_w2
+[2025-10-18 04:40:21,890][02528] Heartbeat connected on RolloutWorker_w4
+[2025-10-18 04:40:21,895][02528] Heartbeat connected on RolloutWorker_w6
+[2025-10-18 04:40:21,900][02528] Heartbeat connected on RolloutWorker_w3
+[2025-10-18 04:40:21,905][02528] Heartbeat connected on RolloutWorker_w5
+[2025-10-18 04:40:21,909][02528] Heartbeat connected on RolloutWorker_w7
+[2025-10-18 04:40:24,652][04336] No checkpoints found
+[2025-10-18 04:40:24,652][04336] Did not load from checkpoint, starting from scratch!
+[2025-10-18 04:40:24,653][04336] Initialized policy 0 weights for model version 0
+[2025-10-18 04:40:24,658][04336] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-18 04:40:24,658][04336] LearnerWorker_p0 finished initialization!
+[2025-10-18 04:40:24,665][02528] Heartbeat connected on LearnerWorker_p0
+[2025-10-18 04:40:24,846][04349] RunningMeanStd input shape: (3, 72, 128)
+[2025-10-18 04:40:24,848][04349] RunningMeanStd input shape: (1,)
+[2025-10-18 04:40:24,862][04349] ConvEncoder: input_channels=3
+[2025-10-18 04:40:25,013][04349] Conv encoder output size: 512
+[2025-10-18 04:40:25,014][04349] Policy head output size: 512
+[2025-10-18 04:40:25,069][02528] Inference worker 0-0 is ready!
+[2025-10-18 04:40:25,072][02528] All inference workers are ready! Signal rollout workers to start!
+[2025-10-18 04:40:25,361][04355] Doom resolution: 160x120, resize resolution: (128, 72)
+[2025-10-18 04:40:25,374][04357] Doom resolution: 160x120, resize resolution: (128, 72)
+[2025-10-18 04:40:25,404][04350] Doom resolution: 160x120, resize resolution: (128, 72)
+[2025-10-18 04:40:25,430][04353] Doom resolution: 160x120, resize resolution: (128, 72)
+[2025-10-18 04:40:25,461][04356] Doom resolution: 160x120, resize resolution: (128, 72)
+[2025-10-18 04:40:25,459][04352] Doom resolution: 160x120, resize resolution: (128, 72)
+[2025-10-18 04:40:25,465][04354] Doom resolution: 160x120, resize resolution: (128, 72)
+[2025-10-18 04:40:25,468][04351] Doom resolution: 160x120, resize resolution: (128, 72)
+[2025-10-18 04:40:25,679][02528] Fps is (10 sec: nan, 60 sec: nan, 300 sec: nan). Total num frames: 0. Throughput: 0: nan. Samples: 0. Policy #0 lag: (min: -1.0, avg: -1.0, max: -1.0)
+[2025-10-18 04:40:26,453][04356] Decorrelating experience for 0 frames...
+[2025-10-18 04:40:26,834][04356] Decorrelating experience for 32 frames...
+[2025-10-18 04:40:27,044][04355] Decorrelating experience for 0 frames...
+[2025-10-18 04:40:27,049][04357] Decorrelating experience for 0 frames...
+[2025-10-18 04:40:27,053][04350] Decorrelating experience for 0 frames...
+[2025-10-18 04:40:27,959][04356] Decorrelating experience for 64 frames...
+[2025-10-18 04:40:28,198][04352] Decorrelating experience for 0 frames...
+[2025-10-18 04:40:28,202][04357] Decorrelating experience for 32 frames...
+[2025-10-18 04:40:28,204][04353] Decorrelating experience for 0 frames...
+[2025-10-18 04:40:28,205][04355] Decorrelating experience for 32 frames...
+[2025-10-18 04:40:28,210][04350] Decorrelating experience for 32 frames...
+[2025-10-18 04:40:28,959][04357] Decorrelating experience for 64 frames...
+[2025-10-18 04:40:29,717][04352] Decorrelating experience for 32 frames...
+[2025-10-18 04:40:29,721][04353] Decorrelating experience for 32 frames...
+[2025-10-18 04:40:29,737][04351] Decorrelating experience for 0 frames...
+[2025-10-18 04:40:29,769][04356] Decorrelating experience for 96 frames...
+[2025-10-18 04:40:29,769][04355] Decorrelating experience for 64 frames...
+[2025-10-18 04:40:29,868][04357] Decorrelating experience for 96 frames...
+[2025-10-18 04:40:30,679][02528] Fps is (10 sec: 0.0, 60 sec: 0.0, 300 sec: 0.0). Total num frames: 0. Throughput: 0: 0.0. Samples: 0. Policy #0 lag: (min: -1.0, avg: -1.0, max: -1.0)
+[2025-10-18 04:40:30,857][04351] Decorrelating experience for 32 frames...
+[2025-10-18 04:40:30,899][04350] Decorrelating experience for 64 frames...
+[2025-10-18 04:40:31,083][04355] Decorrelating experience for 96 frames...
+[2025-10-18 04:40:31,423][04352] Decorrelating experience for 64 frames...
+[2025-10-18 04:40:32,420][04354] Decorrelating experience for 0 frames...
+[2025-10-18 04:40:32,560][04353] Decorrelating experience for 64 frames...
+[2025-10-18 04:40:32,867][04350] Decorrelating experience for 96 frames...
+[2025-10-18 04:40:34,658][04352] Decorrelating experience for 96 frames...
+[2025-10-18 04:40:34,946][04353] Decorrelating experience for 96 frames...
+[2025-10-18 04:40:34,960][04351] Decorrelating experience for 64 frames...
+[2025-10-18 04:40:35,679][02528] Fps is (10 sec: 0.0, 60 sec: 0.0, 300 sec: 0.0). Total num frames: 0. Throughput: 0: 77.4. Samples: 774. Policy #0 lag: (min: -1.0, avg: -1.0, max: -1.0)
+[2025-10-18 04:40:35,699][02528] Avg episode reward: [(0, '-0.483')]
+[2025-10-18 04:40:38,352][04354] Decorrelating experience for 32 frames...
+[2025-10-18 04:40:40,679][02528] Fps is (10 sec: 819.2, 60 sec: 546.1, 300 sec: 546.1). Total num frames: 8192. Throughput: 0: 173.7. Samples: 2606. Policy #0 lag: (min: 0.0, avg: 0.0, max: 0.0)
+[2025-10-18 04:40:40,681][02528] Avg episode reward: [(0, '-0.469')]
+[2025-10-18 04:40:41,513][04351] Decorrelating experience for 96 frames...
+[2025-10-18 04:40:42,326][04354] Decorrelating experience for 64 frames...
+[2025-10-18 04:40:44,510][04354] Decorrelating experience for 96 frames...
+[2025-10-18 04:40:45,679][02528] Fps is (10 sec: 2048.0, 60 sec: 1024.0, 300 sec: 1024.0). Total num frames: 20480. Throughput: 0: 209.1. Samples: 4182. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:40:45,686][02528] Avg episode reward: [(0, '0.054')]
+[2025-10-18 04:40:50,509][04349] Updated weights for policy 0, policy_version 10 (0.0034)
+[2025-10-18 04:40:50,679][02528] Fps is (10 sec: 3276.8, 60 sec: 1638.4, 300 sec: 1638.4). Total num frames: 40960. Throughput: 0: 380.6. Samples: 9516. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:40:50,684][02528] Avg episode reward: [(0, '0.065')]
+[2025-10-18 04:40:55,679][02528] Fps is (10 sec: 3276.9, 60 sec: 1774.9, 300 sec: 1774.9). Total num frames: 53248. Throughput: 0: 466.9. Samples: 14006. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:40:55,684][02528] Avg episode reward: [(0, '0.751')]
+[2025-10-18 04:41:00,679][02528] Fps is (10 sec: 2457.6, 60 sec: 1872.5, 300 sec: 1872.5). Total num frames: 65536. Throughput: 0: 449.1. Samples: 15720. Policy #0 lag: (min: 0.0, avg: 0.4, max: 2.0)
+[2025-10-18 04:41:00,684][02528] Avg episode reward: [(0, '0.668')]
+[2025-10-18 04:41:04,689][04349] Updated weights for policy 0, policy_version 20 (0.0023)
+[2025-10-18 04:41:05,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2048.0, 300 sec: 2048.0). Total num frames: 81920. Throughput: 0: 515.0. Samples: 20598. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:41:05,683][02528] Avg episode reward: [(0, '0.633')]
+[2025-10-18 04:41:10,679][02528] Fps is (10 sec: 3276.8, 60 sec: 2184.5, 300 sec: 2184.5). Total num frames: 98304. Throughput: 0: 567.9. Samples: 25556. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:41:10,685][02528] Avg episode reward: [(0, '1.366')]
+[2025-10-18 04:41:15,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2211.8, 300 sec: 2211.8). Total num frames: 110592. Throughput: 0: 604.7. Samples: 27210. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:41:15,686][02528] Avg episode reward: [(0, '2.481')]
+[2025-10-18 04:41:15,693][04336] Saving new best policy, reward=2.481!
+[2025-10-18 04:41:19,026][04349] Updated weights for policy 0, policy_version 30 (0.0020)
+[2025-10-18 04:41:20,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2308.7, 300 sec: 2308.7). Total num frames: 126976. Throughput: 0: 676.2. Samples: 31204. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:41:20,687][02528] Avg episode reward: [(0, '2.203')]
+[2025-10-18 04:41:25,679][02528] Fps is (10 sec: 3276.9, 60 sec: 2389.3, 300 sec: 2389.3). Total num frames: 143360. Throughput: 0: 751.9. Samples: 36442. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:41:25,681][02528] Avg episode reward: [(0, '2.994')]
+[2025-10-18 04:41:25,688][04336] Saving new best policy, reward=2.994!
+[2025-10-18 04:41:30,681][02528] Fps is (10 sec: 2866.6, 60 sec: 2594.0, 300 sec: 2394.5). Total num frames: 155648. Throughput: 0: 763.1. Samples: 38522. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:41:30,685][02528] Avg episode reward: [(0, '1.620')]
+[2025-10-18 04:41:33,916][04349] Updated weights for policy 0, policy_version 40 (0.0018)
+[2025-10-18 04:41:35,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2798.9, 300 sec: 2399.1). Total num frames: 167936. Throughput: 0: 718.3. Samples: 41838. Policy #0 lag: (min: 0.0, avg: 0.7, max: 2.0)
+[2025-10-18 04:41:35,682][02528] Avg episode reward: [(0, '3.186')]
+[2025-10-18 04:41:35,685][04336] Saving new best policy, reward=3.186!
+[2025-10-18 04:41:40,679][02528] Fps is (10 sec: 2458.1, 60 sec: 2867.2, 300 sec: 2403.0). Total num frames: 180224. Throughput: 0: 710.9. Samples: 45996. Policy #0 lag: (min: 0.0, avg: 0.7, max: 2.0)
+[2025-10-18 04:41:40,682][02528] Avg episode reward: [(0, '1.682')]
+[2025-10-18 04:41:45,681][02528] Fps is (10 sec: 2866.6, 60 sec: 2935.4, 300 sec: 2457.5). Total num frames: 196608. Throughput: 0: 726.9. Samples: 48432. Policy #0 lag: (min: 0.0, avg: 0.4, max: 2.0)
+[2025-10-18 04:41:45,685][02528] Avg episode reward: [(0, '6.078')]
+[2025-10-18 04:41:45,694][04336] Saving new best policy, reward=6.078!
+[2025-10-18 04:41:49,968][04349] Updated weights for policy 0, policy_version 50 (0.0027)
+[2025-10-18 04:41:50,679][02528] Fps is (10 sec: 2457.7, 60 sec: 2730.7, 300 sec: 2409.4). Total num frames: 204800. Throughput: 0: 679.4. Samples: 51172. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:41:50,687][02528] Avg episode reward: [(0, '2.947')]
+[2025-10-18 04:41:55,679][02528] Fps is (10 sec: 2458.1, 60 sec: 2798.9, 300 sec: 2457.6). Total num frames: 221184. Throughput: 0: 670.8. Samples: 55740. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:41:55,684][02528] Avg episode reward: [(0, '3.728')]
+[2025-10-18 04:41:55,694][04336] Saving train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000054_221184.pth...
+[2025-10-18 04:42:00,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2798.9, 300 sec: 2457.6). Total num frames: 233472. Throughput: 0: 675.6. Samples: 57612. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:42:00,685][02528] Avg episode reward: [(0, '2.880')]
+[2025-10-18 04:42:04,740][04349] Updated weights for policy 0, policy_version 60 (0.0025)
+[2025-10-18 04:42:05,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2730.7, 300 sec: 2457.6). Total num frames: 245760. Throughput: 0: 678.6. Samples: 61742. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:42:05,684][02528] Avg episode reward: [(0, '2.537')]
+[2025-10-18 04:42:10,679][02528] Fps is (10 sec: 2457.5, 60 sec: 2662.4, 300 sec: 2457.6). Total num frames: 258048. Throughput: 0: 644.1. Samples: 65428. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:42:10,684][02528] Avg episode reward: [(0, '3.734')]
+[2025-10-18 04:42:15,679][02528] Fps is (10 sec: 3276.8, 60 sec: 2798.9, 300 sec: 2532.1). Total num frames: 278528. Throughput: 0: 656.3. Samples: 68052. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:42:15,687][02528] Avg episode reward: [(0, '2.063')]
+[2025-10-18 04:42:17,887][04349] Updated weights for policy 0, policy_version 70 (0.0015)
+[2025-10-18 04:42:20,679][02528] Fps is (10 sec: 3276.9, 60 sec: 2730.7, 300 sec: 2528.8). Total num frames: 290816. Throughput: 0: 692.8. Samples: 73014. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:42:20,687][02528] Avg episode reward: [(0, '3.895')]
+[2025-10-18 04:42:25,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2662.4, 300 sec: 2525.9). Total num frames: 303104. Throughput: 0: 672.5. Samples: 76260. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:42:25,686][02528] Avg episode reward: [(0, '5.384')]
+[2025-10-18 04:42:30,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2730.8, 300 sec: 2555.9). Total num frames: 319488. Throughput: 0: 673.7. Samples: 78748. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:42:30,683][02528] Avg episode reward: [(0, '4.836')]
+[2025-10-18 04:42:32,211][04349] Updated weights for policy 0, policy_version 80 (0.0025)
+[2025-10-18 04:42:35,680][02528] Fps is (10 sec: 3276.4, 60 sec: 2798.9, 300 sec: 2583.6). Total num frames: 335872. Throughput: 0: 730.4. Samples: 84040. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:42:35,716][02528] Avg episode reward: [(0, '2.635')]
+[2025-10-18 04:42:40,680][02528] Fps is (10 sec: 2866.8, 60 sec: 2798.9, 300 sec: 2578.9). Total num frames: 348160. Throughput: 0: 710.0. Samples: 87690. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:42:40,685][02528] Avg episode reward: [(0, '3.053')]
+[2025-10-18 04:42:45,681][02528] Fps is (10 sec: 2457.5, 60 sec: 2730.7, 300 sec: 2574.6). Total num frames: 360448. Throughput: 0: 705.9. Samples: 89380. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:42:45,689][02528] Avg episode reward: [(0, '4.166')]
+[2025-10-18 04:42:46,796][04349] Updated weights for policy 0, policy_version 90 (0.0026)
+[2025-10-18 04:42:50,679][02528] Fps is (10 sec: 3277.3, 60 sec: 2935.5, 300 sec: 2627.1). Total num frames: 380928. Throughput: 0: 729.4. Samples: 94566. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:42:50,685][02528] Avg episode reward: [(0, '3.264')]
+[2025-10-18 04:42:55,679][02528] Fps is (10 sec: 3277.4, 60 sec: 2867.2, 300 sec: 2621.4). Total num frames: 393216. Throughput: 0: 749.9. Samples: 99174. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:42:55,683][02528] Avg episode reward: [(0, '4.159')]
+[2025-10-18 04:43:00,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2867.2, 300 sec: 2616.2). Total num frames: 405504. Throughput: 0: 727.7. Samples: 100798. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:43:00,681][02528] Avg episode reward: [(0, '4.177')]
+[2025-10-18 04:43:01,411][04349] Updated weights for policy 0, policy_version 100 (0.0022)
+[2025-10-18 04:43:05,679][02528] Fps is (10 sec: 2867.1, 60 sec: 2935.5, 300 sec: 2636.8). Total num frames: 421888. Throughput: 0: 717.7. Samples: 105310. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:43:05,685][02528] Avg episode reward: [(0, '3.483')]
+[2025-10-18 04:43:10,680][02528] Fps is (10 sec: 3276.6, 60 sec: 3003.7, 300 sec: 2656.2). Total num frames: 438272. Throughput: 0: 759.3. Samples: 110430. Policy #0 lag: (min: 0.0, avg: 0.8, max: 2.0)
+[2025-10-18 04:43:10,683][02528] Avg episode reward: [(0, '3.799')]
+[2025-10-18 04:43:15,096][04349] Updated weights for policy 0, policy_version 110 (0.0020)
+[2025-10-18 04:43:15,683][02528] Fps is (10 sec: 2866.1, 60 sec: 2867.0, 300 sec: 2650.3). Total num frames: 450560. Throughput: 0: 743.0. Samples: 112184. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:43:15,690][02528] Avg episode reward: [(0, '4.204')]
+[2025-10-18 04:43:20,680][02528] Fps is (10 sec: 2457.4, 60 sec: 2867.1, 300 sec: 2644.8). Total num frames: 462848. Throughput: 0: 710.3. Samples: 116004. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:43:20,689][02528] Avg episode reward: [(0, '3.735')]
+[2025-10-18 04:43:25,679][02528] Fps is (10 sec: 3278.2, 60 sec: 3003.7, 300 sec: 2685.2). Total num frames: 483328. Throughput: 0: 744.8. Samples: 121204. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:43:25,682][02528] Avg episode reward: [(0, '4.039')]
+[2025-10-18 04:43:27,637][04349] Updated weights for policy 0, policy_version 120 (0.0022)
+[2025-10-18 04:43:30,679][02528] Fps is (10 sec: 3277.1, 60 sec: 2935.4, 300 sec: 2679.0). Total num frames: 495616. Throughput: 0: 762.4. Samples: 123686. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:43:30,690][02528] Avg episode reward: [(0, '4.385')]
+[2025-10-18 04:43:35,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2867.3, 300 sec: 2673.2). Total num frames: 507904. Throughput: 0: 720.8. Samples: 127000. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:43:35,695][02528] Avg episode reward: [(0, '3.702')]
+[2025-10-18 04:43:40,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2935.5, 300 sec: 2688.7). Total num frames: 524288. Throughput: 0: 726.7. Samples: 131876. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:43:40,683][02528] Avg episode reward: [(0, '3.777')]
+[2025-10-18 04:43:42,091][04349] Updated weights for policy 0, policy_version 130 (0.0029)
+[2025-10-18 04:43:45,680][02528] Fps is (10 sec: 3276.7, 60 sec: 3003.8, 300 sec: 2703.4). Total num frames: 540672. Throughput: 0: 749.4. Samples: 134522. Policy #0 lag: (min: 0.0, avg: 0.4, max: 2.0)
+[2025-10-18 04:43:45,690][02528] Avg episode reward: [(0, '3.578')]
+[2025-10-18 04:43:50,681][02528] Fps is (10 sec: 2866.7, 60 sec: 2867.1, 300 sec: 2697.3). Total num frames: 552960. Throughput: 0: 738.2. Samples: 138530. Policy #0 lag: (min: 0.0, avg: 0.4, max: 2.0)
+[2025-10-18 04:43:50,705][02528] Avg episode reward: [(0, '2.809')]
+[2025-10-18 04:43:55,679][02528] Fps is (10 sec: 2867.3, 60 sec: 2935.5, 300 sec: 2711.2). Total num frames: 569344. Throughput: 0: 716.8. Samples: 142684. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:43:55,683][02528] Avg episode reward: [(0, '5.054')]
+[2025-10-18 04:43:55,697][04336] Saving train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000139_569344.pth...
+[2025-10-18 04:43:56,644][04349] Updated weights for policy 0, policy_version 140 (0.0018)
+[2025-10-18 04:44:00,679][02528] Fps is (10 sec: 3277.4, 60 sec: 3003.7, 300 sec: 2724.3). Total num frames: 585728. Throughput: 0: 734.9. Samples: 145250. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:44:00,682][02528] Avg episode reward: [(0, '4.423')]
+[2025-10-18 04:44:05,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2935.5, 300 sec: 2718.3). Total num frames: 598016. Throughput: 0: 756.1. Samples: 150026. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:44:05,688][02528] Avg episode reward: [(0, '4.099')]
+[2025-10-18 04:44:10,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2867.2, 300 sec: 2712.5). Total num frames: 610304. Throughput: 0: 711.6. Samples: 153224. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:44:10,683][02528] Avg episode reward: [(0, '3.542')]
+[2025-10-18 04:44:11,075][04349] Updated weights for policy 0, policy_version 150 (0.0021)
+[2025-10-18 04:44:15,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2935.7, 300 sec: 2724.7). Total num frames: 626688. Throughput: 0: 716.1. Samples: 155912. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:44:15,685][02528] Avg episode reward: [(0, '3.985')]
+[2025-10-18 04:44:20,681][02528] Fps is (10 sec: 3276.1, 60 sec: 3003.7, 300 sec: 2736.5). Total num frames: 643072. Throughput: 0: 760.1. Samples: 161204. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:44:20,689][02528] Avg episode reward: [(0, '3.892')]
+[2025-10-18 04:44:24,449][04349] Updated weights for policy 0, policy_version 160 (0.0023)
+[2025-10-18 04:44:25,679][02528] Fps is (10 sec: 2867.3, 60 sec: 2867.2, 300 sec: 2730.7). Total num frames: 655360. Throughput: 0: 730.9. Samples: 164764. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:44:25,681][02528] Avg episode reward: [(0, '4.474')]
+[2025-10-18 04:44:30,679][02528] Fps is (10 sec: 2867.8, 60 sec: 2935.5, 300 sec: 2741.8). Total num frames: 671744. Throughput: 0: 713.8. Samples: 166644. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:44:30,692][02528] Avg episode reward: [(0, '4.383')]
+[2025-10-18 04:44:35,679][02528] Fps is (10 sec: 3276.8, 60 sec: 3003.7, 300 sec: 2752.5). Total num frames: 688128. Throughput: 0: 742.1. Samples: 171922. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:44:35,687][02528] Avg episode reward: [(0, '5.633')]
+[2025-10-18 04:44:37,105][04349] Updated weights for policy 0, policy_version 170 (0.0028)
+[2025-10-18 04:44:40,680][02528] Fps is (10 sec: 2866.9, 60 sec: 2935.4, 300 sec: 2746.7). Total num frames: 700416. Throughput: 0: 744.0. Samples: 176164. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:44:40,686][02528] Avg episode reward: [(0, '4.093')]
+[2025-10-18 04:44:45,679][02528] Fps is (10 sec: 2457.5, 60 sec: 2867.2, 300 sec: 2741.2). Total num frames: 712704. Throughput: 0: 722.5. Samples: 177764. Policy #0 lag: (min: 0.0, avg: 0.4, max: 2.0)
+[2025-10-18 04:44:45,684][02528] Avg episode reward: [(0, '3.497')]
+[2025-10-18 04:44:50,679][02528] Fps is (10 sec: 3277.2, 60 sec: 3003.8, 300 sec: 2766.7). Total num frames: 733184. Throughput: 0: 724.6. Samples: 182632. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:44:50,685][02528] Avg episode reward: [(0, '5.862')]
+[2025-10-18 04:44:51,796][04349] Updated weights for policy 0, policy_version 180 (0.0025)
+[2025-10-18 04:44:55,679][02528] Fps is (10 sec: 3277.0, 60 sec: 2935.5, 300 sec: 2761.0). Total num frames: 745472. Throughput: 0: 766.6. Samples: 187720. Policy #0 lag: (min: 0.0, avg: 0.7, max: 1.0)
+[2025-10-18 04:44:55,683][02528] Avg episode reward: [(0, '6.190')]
+[2025-10-18 04:44:55,708][04336] Saving new best policy, reward=6.190!
+[2025-10-18 04:45:00,680][02528] Fps is (10 sec: 2457.2, 60 sec: 2867.1, 300 sec: 2755.5). Total num frames: 757760. Throughput: 0: 741.4. Samples: 189274. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:45:00,688][02528] Avg episode reward: [(0, '4.443')]
+[2025-10-18 04:45:05,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2935.5, 300 sec: 2764.8). Total num frames: 774144. Throughput: 0: 714.6. Samples: 193358. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:45:05,684][02528] Avg episode reward: [(0, '4.686')]
+[2025-10-18 04:45:06,107][04349] Updated weights for policy 0, policy_version 190 (0.0017)
+[2025-10-18 04:45:10,679][02528] Fps is (10 sec: 3277.3, 60 sec: 3003.7, 300 sec: 2773.8). Total num frames: 790528. Throughput: 0: 749.0. Samples: 198470. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:45:10,686][02528] Avg episode reward: [(0, '5.912')]
+[2025-10-18 04:45:15,680][02528] Fps is (10 sec: 2866.8, 60 sec: 2935.4, 300 sec: 2768.3). Total num frames: 802816. Throughput: 0: 755.6. Samples: 200646. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:45:15,691][02528] Avg episode reward: [(0, '4.293')]
+[2025-10-18 04:45:20,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2867.3, 300 sec: 2763.1). Total num frames: 815104. Throughput: 0: 703.4. Samples: 203574. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:45:20,684][02528] Avg episode reward: [(0, '5.096')]
+[2025-10-18 04:45:21,239][04349] Updated weights for policy 0, policy_version 200 (0.0028)
+[2025-10-18 04:45:25,679][02528] Fps is (10 sec: 2867.6, 60 sec: 2935.5, 300 sec: 2818.6). Total num frames: 831488. Throughput: 0: 726.3. Samples: 208848. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:45:25,683][02528] Avg episode reward: [(0, '4.874')]
+[2025-10-18 04:45:30,680][02528] Fps is (10 sec: 3276.3, 60 sec: 2935.4, 300 sec: 2874.1). Total num frames: 847872. Throughput: 0: 749.2. Samples: 211480. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:45:30,686][02528] Avg episode reward: [(0, '4.638')]
+[2025-10-18 04:45:35,617][04349] Updated weights for policy 0, policy_version 210 (0.0025)
+[2025-10-18 04:45:35,681][02528] Fps is (10 sec: 2866.7, 60 sec: 2867.1, 300 sec: 2888.0). Total num frames: 860160. Throughput: 0: 712.7. Samples: 214706. Policy #0 lag: (min: 0.0, avg: 0.4, max: 2.0)
+[2025-10-18 04:45:35,686][02528] Avg episode reward: [(0, '4.617')]
+[2025-10-18 04:45:40,679][02528] Fps is (10 sec: 2457.9, 60 sec: 2867.2, 300 sec: 2888.0). Total num frames: 872448. Throughput: 0: 699.2. Samples: 219186. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:45:40,683][02528] Avg episode reward: [(0, '5.419')]
+[2025-10-18 04:45:45,679][02528] Fps is (10 sec: 3277.3, 60 sec: 3003.8, 300 sec: 2888.0). Total num frames: 892928. Throughput: 0: 721.9. Samples: 221758. Policy #0 lag: (min: 0.0, avg: 0.3, max: 1.0)
+[2025-10-18 04:45:45,687][02528] Avg episode reward: [(0, '5.710')]
+[2025-10-18 04:45:48,814][04349] Updated weights for policy 0, policy_version 220 (0.0024)
+[2025-10-18 04:45:50,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2798.9, 300 sec: 2874.1). Total num frames: 901120. Throughput: 0: 722.4. Samples: 225868. Policy #0 lag: (min: 0.0, avg: 0.4, max: 2.0)
+[2025-10-18 04:45:50,685][02528] Avg episode reward: [(0, '3.929')]
+[2025-10-18 04:45:55,683][02528] Fps is (10 sec: 2456.5, 60 sec: 2867.0, 300 sec: 2888.0). Total num frames: 917504. Throughput: 0: 691.8. Samples: 229602. Policy #0 lag: (min: 0.0, avg: 0.7, max: 2.0)
+[2025-10-18 04:45:55,686][02528] Avg episode reward: [(0, '4.380')]
+[2025-10-18 04:45:55,702][04336] Saving train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000224_917504.pth...
+[2025-10-18 04:45:55,920][04336] Removing train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000054_221184.pth
+[2025-10-18 04:46:00,679][02528] Fps is (10 sec: 3276.8, 60 sec: 2935.5, 300 sec: 2888.0). Total num frames: 933888. Throughput: 0: 696.2. Samples: 231974. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:46:00,691][02528] Avg episode reward: [(0, '5.847')]
+[2025-10-18 04:46:03,045][04349] Updated weights for policy 0, policy_version 230 (0.0016)
+[2025-10-18 04:46:05,682][02528] Fps is (10 sec: 2867.5, 60 sec: 2867.0, 300 sec: 2874.1). Total num frames: 946176. Throughput: 0: 736.3. Samples: 236710. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:46:05,689][02528] Avg episode reward: [(0, '5.913')]
+[2025-10-18 04:46:10,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2798.9, 300 sec: 2874.1). Total num frames: 958464. Throughput: 0: 689.3. Samples: 239868. Policy #0 lag: (min: 0.0, avg: 0.3, max: 1.0)
+[2025-10-18 04:46:10,682][02528] Avg episode reward: [(0, '6.065')]
+[2025-10-18 04:46:15,679][02528] Fps is (10 sec: 2868.2, 60 sec: 2867.3, 300 sec: 2874.1). Total num frames: 974848. Throughput: 0: 682.8. Samples: 242206. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:46:15,683][02528] Avg episode reward: [(0, '5.410')]
+[2025-10-18 04:46:17,725][04349] Updated weights for policy 0, policy_version 240 (0.0014)
+[2025-10-18 04:46:20,679][02528] Fps is (10 sec: 3276.7, 60 sec: 2935.4, 300 sec: 2874.1). Total num frames: 991232. Throughput: 0: 725.0. Samples: 247332. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:46:20,684][02528] Avg episode reward: [(0, '4.203')]
+[2025-10-18 04:46:25,680][02528] Fps is (10 sec: 2867.0, 60 sec: 2867.2, 300 sec: 2874.2). Total num frames: 1003520. Throughput: 0: 711.2. Samples: 251192. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:46:25,683][02528] Avg episode reward: [(0, '7.632')]
+[2025-10-18 04:46:25,689][04336] Saving new best policy, reward=7.632!
+[2025-10-18 04:46:30,680][02528] Fps is (10 sec: 2048.0, 60 sec: 2730.7, 300 sec: 2860.3). Total num frames: 1011712. Throughput: 0: 688.5. Samples: 252742. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:46:30,687][02528] Avg episode reward: [(0, '4.357')]
+[2025-10-18 04:46:33,058][04349] Updated weights for policy 0, policy_version 250 (0.0020)
+[2025-10-18 04:46:35,680][02528] Fps is (10 sec: 2867.1, 60 sec: 2867.2, 300 sec: 2888.0). Total num frames: 1032192. Throughput: 0: 701.9. Samples: 257456. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:46:35,686][02528] Avg episode reward: [(0, '4.018')]
+[2025-10-18 04:46:40,679][02528] Fps is (10 sec: 3277.0, 60 sec: 2867.2, 300 sec: 2874.2). Total num frames: 1044480. Throughput: 0: 719.5. Samples: 261978. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:46:40,682][02528] Avg episode reward: [(0, '5.881')]
+[2025-10-18 04:46:45,679][02528] Fps is (10 sec: 2457.8, 60 sec: 2730.7, 300 sec: 2888.0). Total num frames: 1056768. Throughput: 0: 703.2. Samples: 263616. Policy #0 lag: (min: 0.0, avg: 0.6, max: 1.0)
+[2025-10-18 04:46:45,684][02528] Avg episode reward: [(0, '5.979')]
+[2025-10-18 04:46:47,723][04349] Updated weights for policy 0, policy_version 260 (0.0020)
+[2025-10-18 04:46:50,680][02528] Fps is (10 sec: 2866.9, 60 sec: 2867.2, 300 sec: 2888.0). Total num frames: 1073152. Throughput: 0: 698.2. Samples: 268128. Policy #0 lag: (min: 0.0, avg: 0.7, max: 2.0)
+[2025-10-18 04:46:50,684][02528] Avg episode reward: [(0, '6.243')]
+[2025-10-18 04:46:55,679][02528] Fps is (10 sec: 3686.2, 60 sec: 2935.7, 300 sec: 2915.8). Total num frames: 1093632. Throughput: 0: 747.6. Samples: 273510. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:46:55,685][02528] Avg episode reward: [(0, '6.197')]
+[2025-10-18 04:47:00,679][02528] Fps is (10 sec: 2867.5, 60 sec: 2798.9, 300 sec: 2901.9). Total num frames: 1101824. Throughput: 0: 736.8. Samples: 275360. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:47:00,686][02528] Avg episode reward: [(0, '5.088')]
+[2025-10-18 04:47:01,025][04349] Updated weights for policy 0, policy_version 270 (0.0020)
+[2025-10-18 04:47:05,679][02528] Fps is (10 sec: 2048.1, 60 sec: 2799.1, 300 sec: 2901.9). Total num frames: 1114112. Throughput: 0: 695.7. Samples: 278638. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:47:05,684][02528] Avg episode reward: [(0, '4.799')]
+[2025-10-18 04:47:10,679][02528] Fps is (10 sec: 3276.8, 60 sec: 2935.5, 300 sec: 2901.9). Total num frames: 1134592. Throughput: 0: 724.2. Samples: 283782. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:47:10,685][02528] Avg episode reward: [(0, '8.581')]
+[2025-10-18 04:47:10,689][04336] Saving new best policy, reward=8.581!
+[2025-10-18 04:47:14,797][04349] Updated weights for policy 0, policy_version 280 (0.0022)
+[2025-10-18 04:47:15,679][02528] Fps is (10 sec: 3276.8, 60 sec: 2867.2, 300 sec: 2901.9). Total num frames: 1146880. Throughput: 0: 747.0. Samples: 286358. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:47:15,689][02528] Avg episode reward: [(0, '5.566')]
+[2025-10-18 04:47:20,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2798.9, 300 sec: 2901.9). Total num frames: 1159168. Throughput: 0: 715.1. Samples: 289634. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:47:20,684][02528] Avg episode reward: [(0, '3.583')]
+[2025-10-18 04:47:25,679][02528] Fps is (10 sec: 2867.1, 60 sec: 2867.2, 300 sec: 2901.9). Total num frames: 1175552. Throughput: 0: 721.2. Samples: 294432. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:47:25,682][02528] Avg episode reward: [(0, '3.364')]
+[2025-10-18 04:47:28,775][04349] Updated weights for policy 0, policy_version 290 (0.0025)
+[2025-10-18 04:47:30,679][02528] Fps is (10 sec: 3276.7, 60 sec: 3003.7, 300 sec: 2901.9). Total num frames: 1191936. Throughput: 0: 742.3. Samples: 297022. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:47:30,700][02528] Avg episode reward: [(0, '7.713')]
+[2025-10-18 04:47:35,679][02528] Fps is (10 sec: 2867.3, 60 sec: 2867.3, 300 sec: 2901.9). Total num frames: 1204224. Throughput: 0: 727.3. Samples: 300854. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:47:35,683][02528] Avg episode reward: [(0, '6.734')]
+[2025-10-18 04:47:40,679][02528] Fps is (10 sec: 2457.7, 60 sec: 2867.2, 300 sec: 2901.9). Total num frames: 1216512. Throughput: 0: 690.6. Samples: 304588. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:47:40,684][02528] Avg episode reward: [(0, '5.367')]
+[2025-10-18 04:47:43,726][04349] Updated weights for policy 0, policy_version 300 (0.0021)
+[2025-10-18 04:47:45,679][02528] Fps is (10 sec: 2867.1, 60 sec: 2935.5, 300 sec: 2888.0). Total num frames: 1232896. Throughput: 0: 708.4. Samples: 307236. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:47:45,687][02528] Avg episode reward: [(0, '6.527')]
+[2025-10-18 04:47:50,679][02528] Fps is (10 sec: 3276.8, 60 sec: 2935.5, 300 sec: 2901.9). Total num frames: 1249280. Throughput: 0: 747.9. Samples: 312294. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:47:50,682][02528] Avg episode reward: [(0, '8.883')]
+[2025-10-18 04:47:50,689][04336] Saving new best policy, reward=8.883!
+[2025-10-18 04:47:55,681][02528] Fps is (10 sec: 2457.1, 60 sec: 2730.6, 300 sec: 2888.0). Total num frames: 1257472. Throughput: 0: 703.1. Samples: 315424. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:47:55,684][02528] Avg episode reward: [(0, '4.580')]
+[2025-10-18 04:47:55,701][04336] Saving train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000307_1257472.pth...
+[2025-10-18 04:47:55,957][04336] Removing train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000139_569344.pth
+[2025-10-18 04:47:58,420][04349] Updated weights for policy 0, policy_version 310 (0.0032)
+[2025-10-18 04:48:00,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2867.2, 300 sec: 2888.0). Total num frames: 1273856. Throughput: 0: 697.5. Samples: 317746. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:48:00,684][02528] Avg episode reward: [(0, '5.529')]
+[2025-10-18 04:48:05,680][02528] Fps is (10 sec: 3277.0, 60 sec: 2935.4, 300 sec: 2888.0). Total num frames: 1290240. Throughput: 0: 735.8. Samples: 322748. Policy #0 lag: (min: 0.0, avg: 0.6, max: 2.0)
+[2025-10-18 04:48:05,691][02528] Avg episode reward: [(0, '6.455')]
+[2025-10-18 04:48:10,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2798.9, 300 sec: 2888.1). Total num frames: 1302528. Throughput: 0: 712.5. Samples: 326494. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:48:10,683][02528] Avg episode reward: [(0, '5.723')]
+[2025-10-18 04:48:13,124][04349] Updated weights for policy 0, policy_version 320 (0.0020)
+[2025-10-18 04:48:15,679][02528] Fps is (10 sec: 2867.6, 60 sec: 2867.2, 300 sec: 2901.9). Total num frames: 1318912. Throughput: 0: 691.6. Samples: 328144. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:48:15,686][02528] Avg episode reward: [(0, '4.236')]
+[2025-10-18 04:48:20,679][02528] Fps is (10 sec: 3276.9, 60 sec: 2935.5, 300 sec: 2888.0). Total num frames: 1335296. Throughput: 0: 719.3. Samples: 333222. Policy #0 lag: (min: 0.0, avg: 0.4, max: 1.0)
+[2025-10-18 04:48:20,684][02528] Avg episode reward: [(0, '5.477')]
+[2025-10-18 04:48:25,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2867.2, 300 sec: 2888.0). Total num frames: 1347584. Throughput: 0: 735.0. Samples: 337662. Policy #0 lag: (min: 0.0, avg: 0.5, max: 2.0)
+[2025-10-18 04:48:25,681][02528] Avg episode reward: [(0, '9.114')]
+[2025-10-18 04:48:25,685][04336] Saving new best policy, reward=9.114!
+[2025-10-18 04:48:26,644][04349] Updated weights for policy 0, policy_version 330 (0.0023)
+[2025-10-18 04:48:30,679][02528] Fps is (10 sec: 2457.6, 60 sec: 2799.0, 300 sec: 2888.0). Total num frames: 1359872. Throughput: 0: 710.6. Samples: 339214. Policy #0 lag: (min: 0.0, avg: 0.5, max: 1.0)
+[2025-10-18 04:48:30,682][02528] Avg episode reward: [(0, '6.420')]
+[2025-10-18 04:48:35,679][02528] Fps is (10 sec: 2867.2, 60 sec: 2867.2, 300 sec: 2888.0). Total num frames: 1376256. Throughput: 0: 694.0. Samples: 343526. Policy #0 lag: (min: 0.0, avg: 0.4, max: 2.0)
+[2025-10-18 04:48:35,685][02528] Avg episode reward: [(0, '6.379')]
+[2025-10-18 04:48:36,139][02528] Keyboard interrupt detected in the event loop EvtLoop [Runner_EvtLoop, process=main process 2528], exiting...
+[2025-10-18 04:48:36,148][04336] Stopping Batcher_0...
+[2025-10-18 04:48:36,151][04336] Loop batcher_evt_loop terminating...
+[2025-10-18 04:48:36,155][04336] Saving train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000336_1376256.pth...
+[2025-10-18 04:48:36,150][02528] Runner profile tree view:
+main_loop: 514.2599
+[2025-10-18 04:48:36,168][02528] Collected {0: 1376256}, FPS: 2676.2
+[2025-10-18 04:48:36,244][04349] Weights refcount: 2 0
+[2025-10-18 04:48:36,249][04349] Stopping InferenceWorker_p0-w0...
+[2025-10-18 04:48:36,252][04349] Loop inference_proc0-0_evt_loop terminating...
+[2025-10-18 04:48:36,379][04336] Removing train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000224_917504.pth
+[2025-10-18 04:48:36,355][04351] EvtLoop [rollout_proc0_evt_loop, process=rollout_proc0] unhandled exception in slot='advance_rollouts' connected to emitter=Emitter(object_id='InferenceWorker_p0-w0', signal_name='advance0'), args=(0, 0)
+Traceback (most recent call last):
+  File "/usr/local/lib/python3.12/dist-packages/signal_slot/signal_slot.py", line 355, in _process_signal
+    slot_callable(*args)
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/sampling/rollout_worker.py", line 241, in advance_rollouts
+    complete_rollouts, episodic_stats = runner.advance_rollouts(policy_id, self.timing)
+                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/sampling/non_batched_sampling.py", line 634, in advance_rollouts
+    new_obs, rewards, terminated, truncated, infos = e.step(actions)
+                                                     ^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/gymnasium/core.py", line 461, in step
+    return self.env.step(action)
+           ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/utils/make_env.py", line 129, in step
+    obs, rew, terminated, truncated, info = self.env.step(action)
+                                            ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/utils/make_env.py", line 115, in step
+    obs, rew, terminated, truncated, info = self.env.step(action)
+                                            ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/gymnasium/core.py", line 555, in step
+    observation, reward, terminated, truncated, info = self.env.step(action)
+                                                       ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/gymnasium/core.py", line 522, in step
+    observation, reward, terminated, truncated, info = self.env.step(action)
+                                                       ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/envs/env_wrappers.py", line 86, in step
+    obs, reward, terminated, truncated, info = self.env.step(action)
+                                               ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/gymnasium/core.py", line 461, in step
+    return self.env.step(action)
+           ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sf_examples/vizdoom/doom/wrappers/multiplayer_stats.py", line 54, in step
+    obs, reward, terminated, truncated, info = self.env.step(action)
+                                               ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sf_examples/vizdoom/doom/doom_gym.py", line 452, in step
+    reward = self.game.make_action(actions_flattened, self.skip_frames)
+             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+vizdoom.vizdoom.SignalException: Signal SIGINT received. ViZDoom instance has been closed.
+[2025-10-18 04:48:36,417][04336] Stopping LearnerWorker_p0...
+[2025-10-18 04:48:36,419][04336] Loop learner_proc0_evt_loop terminating...
+[2025-10-18 04:48:36,413][04351] Unhandled exception Signal SIGINT received. ViZDoom instance has been closed. in evt loop rollout_proc0_evt_loop
+[2025-10-18 04:48:36,478][04356] EvtLoop [rollout_proc6_evt_loop, process=rollout_proc6] unhandled exception in slot='advance_rollouts' connected to emitter=Emitter(object_id='InferenceWorker_p0-w0', signal_name='advance6'), args=(0, 0)
+Traceback (most recent call last):
+  File "/usr/local/lib/python3.12/dist-packages/signal_slot/signal_slot.py", line 355, in _process_signal
+    slot_callable(*args)
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/sampling/rollout_worker.py", line 241, in advance_rollouts
+    complete_rollouts, episodic_stats = runner.advance_rollouts(policy_id, self.timing)
+                                        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/sampling/non_batched_sampling.py", line 634, in advance_rollouts
+    new_obs, rewards, terminated, truncated, infos = e.step(actions)
+                                                     ^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/gymnasium/core.py", line 461, in step
+    return self.env.step(action)
+           ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/utils/make_env.py", line 129, in step
+    obs, rew, terminated, truncated, info = self.env.step(action)
+                                            ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/utils/make_env.py", line 115, in step
+    obs, rew, terminated, truncated, info = self.env.step(action)
+                                            ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/gymnasium/core.py", line 555, in step
+    observation, reward, terminated, truncated, info = self.env.step(action)
+                                                       ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/gymnasium/core.py", line 522, in step
+    observation, reward, terminated, truncated, info = self.env.step(action)
+                                                       ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/envs/env_wrappers.py", line 86, in step
+    obs, reward, terminated, truncated, info = self.env.step(action)
+                                               ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/gymnasium/core.py", line 461, in step
+    return self.env.step(action)
+           ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sf_examples/vizdoom/doom/wrappers/multiplayer_stats.py", line 54, in step
+    obs, reward, terminated, truncated, info = self.env.step(action)
+                                               ^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/sf_examples/vizdoom/doom/doom_gym.py", line 452, in step
+    reward = self.game.make_action(actions_flattened, self.skip_frames)
+             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+vizdoom.vizdoom.SignalException: Signal SIGINT received. ViZDoom instance has been closed.
+[2025-10-18 04:48:36,537][04356] Unhandled exception Signal SIGINT received. ViZDoom instance has been closed. in evt loop rollout_proc6_evt_loop
+[2025-10-18 18:37:47,967][01690] Saving configuration to train_dir/deadly_corridor_experiment/config.json...
+[2025-10-18 18:37:48,722][01690] Rollout worker 0 uses device cpu
+[2025-10-18 18:37:48,850][01690] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-18 18:37:48,852][01690] InferenceWorker_p0-w0: min num requests: 1
+[2025-10-18 18:37:48,862][01690] Starting all processes...
+[2025-10-18 18:37:48,866][01690] Starting process learner_proc0
+[2025-10-18 18:37:48,972][01690] Starting all processes...
+[2025-10-18 18:37:48,980][01690] Starting process inference_proc0-0
+[2025-10-18 18:37:48,982][01690] Starting process rollout_proc0
+[2025-10-18 18:37:54,595][02697] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-18 18:37:54,597][02697] Set environment var CUDA_VISIBLE_DEVICES to '0' (GPU indices [0]) for learning process 0
+[2025-10-18 18:37:54,618][02697] Num visible devices: 1
+[2025-10-18 18:37:54,625][02697] Starting seed is not provided
+[2025-10-18 18:37:54,626][02697] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-18 18:37:54,626][02697] Initializing actor-critic model on device cuda:0
+[2025-10-18 18:37:54,627][02697] RunningMeanStd input shape: (3, 72, 128)
+[2025-10-18 18:37:54,630][02697] RunningMeanStd input shape: (1,)
+[2025-10-18 18:37:54,662][02697] ConvEncoder: input_channels=3
+[2025-10-18 18:37:55,082][02704] Worker 0 uses CPU cores [0, 1]
+[2025-10-18 18:37:55,097][02703] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-18 18:37:55,098][02703] Set environment var CUDA_VISIBLE_DEVICES to '0' (GPU indices [0]) for inference process 0
+[2025-10-18 18:37:55,115][02697] Conv encoder output size: 512
+[2025-10-18 18:37:55,115][02703] Num visible devices: 1
+[2025-10-18 18:37:55,116][02697] Policy head output size: 512
+[2025-10-18 18:37:55,159][02697] Created Actor Critic model with architecture:
+[2025-10-18 18:37:55,159][02697] ActorCriticSharedWeights(
+  (obs_normalizer): ObservationNormalizer(
+    (running_mean_std): RunningMeanStdDictInPlace(
+      (running_mean_std): ModuleDict(
+        (obs): RunningMeanStdInPlace()
+      )
+    )
+  )
+  (returns_normalizer): RecursiveScriptModule(original_name=RunningMeanStdInPlace)
+  (encoder): VizdoomEncoder(
+    (basic_encoder): ConvEncoder(
+      (enc): RecursiveScriptModule(
+        original_name=ConvEncoderImpl
+        (conv_head): RecursiveScriptModule(
+          original_name=Sequential
+          (0): RecursiveScriptModule(original_name=Conv2d)
+          (1): RecursiveScriptModule(original_name=ELU)
+          (2): RecursiveScriptModule(original_name=Conv2d)
+          (3): RecursiveScriptModule(original_name=ELU)
+          (4): RecursiveScriptModule(original_name=Conv2d)
+          (5): RecursiveScriptModule(original_name=ELU)
+        )
+        (mlp_layers): RecursiveScriptModule(
+          original_name=Sequential
+          (0): RecursiveScriptModule(original_name=Linear)
+          (1): RecursiveScriptModule(original_name=ELU)
+        )
+      )
+    )
+  )
+  (core): ModelCoreRNN(
+    (core): GRU(512, 512)
+  )
+  (decoder): MlpDecoder(
+    (mlp): Identity()
+  )
+  (critic_linear): Linear(in_features=512, out_features=1, bias=True)
+  (action_parameterization): ActionParameterizationDefault(
+    (distribution_linear): Linear(in_features=512, out_features=11, bias=True)
+  )
+)
+[2025-10-18 18:37:55,418][02697] Using optimizer <class 'torch.optim.adam.Adam'>
+[2025-10-18 18:37:59,948][02697] Loading state from checkpoint train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000336_1376256.pth...
+[2025-10-18 18:38:04,136][02697] Could not load from checkpoint, attempt 0
+Traceback (most recent call last):
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/learning/learner.py", line 281, in load_checkpoint
+    checkpoint_dict = torch.load(latest_checkpoint, map_location=device)
+                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/torch/serialization.py", line 1529, in load
+    raise pickle.UnpicklingError(_get_wo_message(str(e))) from None
+_pickle.UnpicklingError: Weights only load failed. This file can still be loaded, to do so you have two options, do those steps only if you trust the source of the checkpoint. 
+	(1) In PyTorch 2.6, we changed the default value of the `weights_only` argument in `torch.load` from `False` to `True`. Re-running `torch.load` with `weights_only` set to `False` will likely succeed, but it can result in arbitrary code execution. Do it only if you got the file from a trusted source.
+	(2) Alternatively, to load with `weights_only=True` please check the recommended steps in the following error message.
+	WeightsUnpickler error: Unsupported global: GLOBAL numpy.core.multiarray.scalar was not an allowed global by default. Please use `torch.serialization.add_safe_globals([numpy.core.multiarray.scalar])` or the `torch.serialization.safe_globals([numpy.core.multiarray.scalar])` context manager to allowlist this global if you trust this class/function.
+
+Check the documentation of torch.load to learn more about types accepted by default with weights_only https://pytorch.org/docs/stable/generated/torch.load.html.
+[2025-10-18 18:38:04,140][02697] Loading state from checkpoint train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000336_1376256.pth...
+[2025-10-18 18:38:04,146][02697] Could not load from checkpoint, attempt 1
+Traceback (most recent call last):
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/learning/learner.py", line 281, in load_checkpoint
+    checkpoint_dict = torch.load(latest_checkpoint, map_location=device)
+                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/torch/serialization.py", line 1529, in load
+    raise pickle.UnpicklingError(_get_wo_message(str(e))) from None
+_pickle.UnpicklingError: Weights only load failed. This file can still be loaded, to do so you have two options, do those steps only if you trust the source of the checkpoint. 
+	(1) In PyTorch 2.6, we changed the default value of the `weights_only` argument in `torch.load` from `False` to `True`. Re-running `torch.load` with `weights_only` set to `False` will likely succeed, but it can result in arbitrary code execution. Do it only if you got the file from a trusted source.
+	(2) Alternatively, to load with `weights_only=True` please check the recommended steps in the following error message.
+	WeightsUnpickler error: Unsupported global: GLOBAL numpy.core.multiarray.scalar was not an allowed global by default. Please use `torch.serialization.add_safe_globals([numpy.core.multiarray.scalar])` or the `torch.serialization.safe_globals([numpy.core.multiarray.scalar])` context manager to allowlist this global if you trust this class/function.
+
+Check the documentation of torch.load to learn more about types accepted by default with weights_only https://pytorch.org/docs/stable/generated/torch.load.html.
+[2025-10-18 18:38:04,147][02697] Loading state from checkpoint train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000336_1376256.pth...
+[2025-10-18 18:38:04,154][02697] Could not load from checkpoint, attempt 2
+Traceback (most recent call last):
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/learning/learner.py", line 281, in load_checkpoint
+    checkpoint_dict = torch.load(latest_checkpoint, map_location=device)
+                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/torch/serialization.py", line 1529, in load
+    raise pickle.UnpicklingError(_get_wo_message(str(e))) from None
+_pickle.UnpicklingError: Weights only load failed. This file can still be loaded, to do so you have two options, do those steps only if you trust the source of the checkpoint. 
+	(1) In PyTorch 2.6, we changed the default value of the `weights_only` argument in `torch.load` from `False` to `True`. Re-running `torch.load` with `weights_only` set to `False` will likely succeed, but it can result in arbitrary code execution. Do it only if you got the file from a trusted source.
+	(2) Alternatively, to load with `weights_only=True` please check the recommended steps in the following error message.
+	WeightsUnpickler error: Unsupported global: GLOBAL numpy.core.multiarray.scalar was not an allowed global by default. Please use `torch.serialization.add_safe_globals([numpy.core.multiarray.scalar])` or the `torch.serialization.safe_globals([numpy.core.multiarray.scalar])` context manager to allowlist this global if you trust this class/function.
+
+Check the documentation of torch.load to learn more about types accepted by default with weights_only https://pytorch.org/docs/stable/generated/torch.load.html.
+[2025-10-18 18:38:04,155][02697] Did not load from checkpoint, starting from scratch!
+[2025-10-18 18:38:04,155][02697] Initialized policy 0 weights for model version 0
+[2025-10-18 18:38:04,160][02697] LearnerWorker_p0 finished initialization!
+[2025-10-18 18:38:04,165][02697] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-18 18:38:04,335][02703] RunningMeanStd input shape: (3, 72, 128)
+[2025-10-18 18:38:04,337][02703] RunningMeanStd input shape: (1,)
+[2025-10-18 18:38:04,354][02703] ConvEncoder: input_channels=3
+[2025-10-18 18:38:04,494][02703] Conv encoder output size: 512
+[2025-10-18 18:38:04,495][02703] Policy head output size: 512
+[2025-10-18 18:38:04,541][01690] Inference worker 0-0 is ready!
+[2025-10-18 18:38:04,545][01690] All inference workers are ready! Signal rollout workers to start!
+[2025-10-18 18:38:04,596][02704] Doom resolution: 160x120, resize resolution: (128, 72)
+[2025-10-18 18:38:05,722][02704] Decorrelating experience for 0 frames...
+[2025-10-18 18:38:06,084][02704] Decorrelating experience for 32 frames...
+[2025-10-18 18:38:06,485][02704] Decorrelating experience for 64 frames...
+[2025-10-18 18:38:06,758][02704] Decorrelating experience for 96 frames...
+[2025-10-18 18:38:07,287][01690] Fps is (10 sec: nan, 60 sec: nan, 300 sec: nan). Total num frames: 0. Throughput: 0: nan. Samples: 0. Policy #0 lag: (min: -1.0, avg: -1.0, max: -1.0)
+[2025-10-18 18:38:08,843][01690] Heartbeat connected on Batcher_0
+[2025-10-18 18:38:08,850][01690] Heartbeat connected on LearnerWorker_p0
+[2025-10-18 18:38:08,861][01690] Heartbeat connected on InferenceWorker_p0-w0
+[2025-10-18 18:38:08,868][01690] Heartbeat connected on RolloutWorker_w0
+[2025-10-18 18:38:12,287][01690] Fps is (10 sec: 0.0, 60 sec: 0.0, 300 sec: 0.0). Total num frames: 0. Throughput: 0: 256.8. Samples: 1284. Policy #0 lag: (min: -1.0, avg: -1.0, max: -1.0)
+[2025-10-18 18:38:12,294][01690] Avg episode reward: [(0, '0.993')]
+[2025-10-18 18:38:12,967][02697] Signal inference workers to stop experience collection...
+[2025-10-18 18:38:12,975][02703] InferenceWorker_p0-w0: stopping experience collection
+[2025-10-18 18:38:13,187][02697] Signal inference workers to resume experience collection...
+[2025-10-18 18:38:13,190][02703] InferenceWorker_p0-w0: resuming experience collection
+[2025-10-18 18:38:17,238][02697] Stopping Batcher_0...
+[2025-10-18 18:38:17,244][02697] Loop batcher_evt_loop terminating...
+[2025-10-18 18:38:17,239][01690] Component Batcher_0 stopped!
+[2025-10-18 18:38:17,249][02697] Saving train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000004_16384.pth...
+[2025-10-18 18:38:17,286][02703] Weights refcount: 2 0
+[2025-10-18 18:38:17,294][02703] Stopping InferenceWorker_p0-w0...
+[2025-10-18 18:38:17,295][01690] Component InferenceWorker_p0-w0 stopped!
+[2025-10-18 18:38:17,301][02703] Loop inference_proc0-0_evt_loop terminating...
+[2025-10-18 18:38:17,603][02697] Removing train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000004_16384.pth
+[2025-10-18 18:38:17,615][02704] Stopping RolloutWorker_w0...
+[2025-10-18 18:38:17,618][02697] Saving train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000004_16384.pth...
+[2025-10-18 18:38:17,617][02704] Loop rollout_proc0_evt_loop terminating...
+[2025-10-18 18:38:17,616][01690] Component RolloutWorker_w0 stopped!
+[2025-10-18 18:38:17,981][02697] Removing train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000004_16384.pth
+[2025-10-18 18:38:17,998][02697] Stopping LearnerWorker_p0...
+[2025-10-18 18:38:18,001][02697] Loop learner_proc0_evt_loop terminating...
+[2025-10-18 18:38:18,000][01690] Component LearnerWorker_p0 stopped!
+[2025-10-18 18:38:18,012][01690] Waiting for process learner_proc0 to stop...
+[2025-10-18 18:38:19,658][01690] Waiting for process inference_proc0-0 to join...
+[2025-10-18 18:38:19,665][01690] Waiting for process rollout_proc0 to join...
+[2025-10-18 18:38:19,671][01690] Batcher 0 profile tree view:
+batching: 0.1554, releasing_batches: 0.0031
+[2025-10-18 18:38:19,676][01690] InferenceWorker_p0-w0 profile tree view:
+wait_policy: 0.0000
+  wait_policy_total: 2.9008
+update_model: 0.1376
+  weight_update: 0.0198
+one_step: 0.0033
+  handle_policy_step: 9.1288
+    deserialize: 0.1130, stack: 0.0367, obs_to_device_normalize: 1.8738, forward: 5.9609, send_messages: 0.2287
+    prepare_outputs: 0.6964
+      to_cpu: 0.4763
+[2025-10-18 18:38:19,679][01690] Learner 0 profile tree view:
+misc: 0.0000, prepare_batch: 1.2749
+train: 2.5333
+  epoch_init: 0.0000, minibatch_init: 0.0000, losses_postprocess: 0.0019, kl_divergence: 0.0198, after_optimizer: 0.1426
+  calculate_losses: 0.7245
+    losses_init: 0.0000, forward_head: 0.4491, bptt_initial: 0.1020, tail: 0.0716, advantages_returns: 0.0014, losses: 0.0882
+    bptt: 0.0113
+      bptt_forward_core: 0.0110
+  update: 1.6398
+    clip: 0.0674
+[2025-10-18 18:38:19,682][01690] RolloutWorker_w0 profile tree view:
+wait_for_trajectories: 0.0058, enqueue_policy_requests: 0.3530, env_step: 5.5468, overhead: 0.2136, complete_rollouts: 0.0110
+save_policy_outputs: 0.2790
+  split_output_tensors: 0.1091
+[2025-10-18 18:38:19,693][01690] Loop Runner_EvtLoop terminating...
+[2025-10-18 18:38:19,695][01690] Runner profile tree view:
+main_loop: 30.8336
+[2025-10-18 18:38:19,700][01690] Collected {0: 16384}, FPS: 531.4
+[2025-10-19 00:00:26,666][14521] Saving configuration to train_dir/deadly_corridor_experiment/config.json...
+[2025-10-19 00:00:27,422][14521] Rollout worker 0 uses device cpu
+[2025-10-19 00:00:27,565][14521] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-19 00:00:27,566][14521] InferenceWorker_p0-w0: min num requests: 1
+[2025-10-19 00:00:27,575][14521] Starting all processes...
+[2025-10-19 00:00:27,578][14521] Starting process learner_proc0
+[2025-10-19 00:00:29,749][14521] Starting all processes...
+[2025-10-19 00:00:29,754][14521] Starting process inference_proc0-0
+[2025-10-19 00:00:29,754][14521] Starting process rollout_proc0
+[2025-10-19 00:00:29,767][14634] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-19 00:00:29,770][14634] Set environment var CUDA_VISIBLE_DEVICES to '0' (GPU indices [0]) for learning process 0
+[2025-10-19 00:00:29,794][14634] Num visible devices: 1
+[2025-10-19 00:00:29,802][14634] Starting seed is not provided
+[2025-10-19 00:00:29,802][14634] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-19 00:00:29,803][14634] Initializing actor-critic model on device cuda:0
+[2025-10-19 00:00:29,805][14634] RunningMeanStd input shape: (3, 72, 128)
+[2025-10-19 00:00:29,808][14634] RunningMeanStd input shape: (1,)
+[2025-10-19 00:00:29,843][14634] ConvEncoder: input_channels=3
+[2025-10-19 00:00:30,232][14634] Conv encoder output size: 512
+[2025-10-19 00:00:30,232][14634] Policy head output size: 512
+[2025-10-19 00:00:30,311][14634] Created Actor Critic model with architecture:
+[2025-10-19 00:00:30,312][14634] ActorCriticSharedWeights(
+  (obs_normalizer): ObservationNormalizer(
+    (running_mean_std): RunningMeanStdDictInPlace(
+      (running_mean_std): ModuleDict(
+        (obs): RunningMeanStdInPlace()
+      )
+    )
+  )
+  (returns_normalizer): RecursiveScriptModule(original_name=RunningMeanStdInPlace)
+  (encoder): VizdoomEncoder(
+    (basic_encoder): ConvEncoder(
+      (enc): RecursiveScriptModule(
+        original_name=ConvEncoderImpl
+        (conv_head): RecursiveScriptModule(
+          original_name=Sequential
+          (0): RecursiveScriptModule(original_name=Conv2d)
+          (1): RecursiveScriptModule(original_name=ELU)
+          (2): RecursiveScriptModule(original_name=Conv2d)
+          (3): RecursiveScriptModule(original_name=ELU)
+          (4): RecursiveScriptModule(original_name=Conv2d)
+          (5): RecursiveScriptModule(original_name=ELU)
+        )
+        (mlp_layers): RecursiveScriptModule(
+          original_name=Sequential
+          (0): RecursiveScriptModule(original_name=Linear)
+          (1): RecursiveScriptModule(original_name=ELU)
+        )
+      )
+    )
+  )
+  (core): ModelCoreRNN(
+    (core): GRU(512, 512)
+  )
+  (decoder): MlpDecoder(
+    (mlp): Identity()
+  )
+  (critic_linear): Linear(in_features=512, out_features=1, bias=True)
+  (action_parameterization): ActionParameterizationDefault(
+    (distribution_linear): Linear(in_features=512, out_features=11, bias=True)
+  )
+)
+[2025-10-19 00:00:30,816][14634] Using optimizer <class 'torch.optim.adam.Adam'>
+[2025-10-19 00:00:34,481][14652] Worker 0 uses CPU cores [0, 1]
+[2025-10-19 00:00:34,723][14651] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-19 00:00:34,730][14651] Set environment var CUDA_VISIBLE_DEVICES to '0' (GPU indices [0]) for inference process 0
+[2025-10-19 00:00:34,759][14651] Num visible devices: 1
+[2025-10-19 00:00:38,691][14634] Loading state from checkpoint train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000336_1376256.pth...
+[2025-10-19 00:00:40,739][14634] Could not load from checkpoint, attempt 0
+Traceback (most recent call last):
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/learning/learner.py", line 281, in load_checkpoint
+    checkpoint_dict = torch.load(latest_checkpoint, map_location=device)
+                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/torch/serialization.py", line 1529, in load
+    raise pickle.UnpicklingError(_get_wo_message(str(e))) from None
+_pickle.UnpicklingError: Weights only load failed. This file can still be loaded, to do so you have two options, do those steps only if you trust the source of the checkpoint. 
+	(1) In PyTorch 2.6, we changed the default value of the `weights_only` argument in `torch.load` from `False` to `True`. Re-running `torch.load` with `weights_only` set to `False` will likely succeed, but it can result in arbitrary code execution. Do it only if you got the file from a trusted source.
+	(2) Alternatively, to load with `weights_only=True` please check the recommended steps in the following error message.
+	WeightsUnpickler error: Unsupported global: GLOBAL numpy.core.multiarray.scalar was not an allowed global by default. Please use `torch.serialization.add_safe_globals([numpy.core.multiarray.scalar])` or the `torch.serialization.safe_globals([numpy.core.multiarray.scalar])` context manager to allowlist this global if you trust this class/function.
+
+Check the documentation of torch.load to learn more about types accepted by default with weights_only https://pytorch.org/docs/stable/generated/torch.load.html.
+[2025-10-19 00:00:40,742][14634] Loading state from checkpoint train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000336_1376256.pth...
+[2025-10-19 00:00:40,781][14634] Could not load from checkpoint, attempt 1
+Traceback (most recent call last):
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/learning/learner.py", line 281, in load_checkpoint
+    checkpoint_dict = torch.load(latest_checkpoint, map_location=device)
+                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/torch/serialization.py", line 1529, in load
+    raise pickle.UnpicklingError(_get_wo_message(str(e))) from None
+_pickle.UnpicklingError: Weights only load failed. This file can still be loaded, to do so you have two options, do those steps only if you trust the source of the checkpoint. 
+	(1) In PyTorch 2.6, we changed the default value of the `weights_only` argument in `torch.load` from `False` to `True`. Re-running `torch.load` with `weights_only` set to `False` will likely succeed, but it can result in arbitrary code execution. Do it only if you got the file from a trusted source.
+	(2) Alternatively, to load with `weights_only=True` please check the recommended steps in the following error message.
+	WeightsUnpickler error: Unsupported global: GLOBAL numpy.core.multiarray.scalar was not an allowed global by default. Please use `torch.serialization.add_safe_globals([numpy.core.multiarray.scalar])` or the `torch.serialization.safe_globals([numpy.core.multiarray.scalar])` context manager to allowlist this global if you trust this class/function.
+
+Check the documentation of torch.load to learn more about types accepted by default with weights_only https://pytorch.org/docs/stable/generated/torch.load.html.
+[2025-10-19 00:00:40,782][14634] Loading state from checkpoint train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000336_1376256.pth...
+[2025-10-19 00:00:40,785][14634] Could not load from checkpoint, attempt 2
+Traceback (most recent call last):
+  File "/usr/local/lib/python3.12/dist-packages/sample_factory/algo/learning/learner.py", line 281, in load_checkpoint
+    checkpoint_dict = torch.load(latest_checkpoint, map_location=device)
+                      ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
+  File "/usr/local/lib/python3.12/dist-packages/torch/serialization.py", line 1529, in load
+    raise pickle.UnpicklingError(_get_wo_message(str(e))) from None
+_pickle.UnpicklingError: Weights only load failed. This file can still be loaded, to do so you have two options, do those steps only if you trust the source of the checkpoint. 
+	(1) In PyTorch 2.6, we changed the default value of the `weights_only` argument in `torch.load` from `False` to `True`. Re-running `torch.load` with `weights_only` set to `False` will likely succeed, but it can result in arbitrary code execution. Do it only if you got the file from a trusted source.
+	(2) Alternatively, to load with `weights_only=True` please check the recommended steps in the following error message.
+	WeightsUnpickler error: Unsupported global: GLOBAL numpy.core.multiarray.scalar was not an allowed global by default. Please use `torch.serialization.add_safe_globals([numpy.core.multiarray.scalar])` or the `torch.serialization.safe_globals([numpy.core.multiarray.scalar])` context manager to allowlist this global if you trust this class/function.
+
+Check the documentation of torch.load to learn more about types accepted by default with weights_only https://pytorch.org/docs/stable/generated/torch.load.html.
+[2025-10-19 00:00:40,786][14634] Did not load from checkpoint, starting from scratch!
+[2025-10-19 00:00:40,786][14634] Initialized policy 0 weights for model version 0
+[2025-10-19 00:00:40,790][14634] LearnerWorker_p0 finished initialization!
+[2025-10-19 00:00:40,792][14634] Using GPUs [0] for process 0 (actually maps to GPUs [0])
+[2025-10-19 00:00:40,905][14651] RunningMeanStd input shape: (3, 72, 128)
+[2025-10-19 00:00:40,906][14651] RunningMeanStd input shape: (1,)
+[2025-10-19 00:00:40,916][14651] ConvEncoder: input_channels=3
+[2025-10-19 00:00:41,006][14651] Conv encoder output size: 512
+[2025-10-19 00:00:41,007][14651] Policy head output size: 512
+[2025-10-19 00:00:41,039][14521] Inference worker 0-0 is ready!
+[2025-10-19 00:00:41,040][14521] All inference workers are ready! Signal rollout workers to start!
+[2025-10-19 00:00:41,069][14521] Fps is (10 sec: nan, 60 sec: nan, 300 sec: nan). Total num frames: 0. Throughput: 0: nan. Samples: 0. Policy #0 lag: (min: -1.0, avg: -1.0, max: -1.0)
+[2025-10-19 00:00:41,077][14652] Doom resolution: 160x120, resize resolution: (128, 72)
+[2025-10-19 00:00:41,830][14652] Decorrelating experience for 0 frames...
+[2025-10-19 00:00:42,060][14652] Decorrelating experience for 32 frames...
+[2025-10-19 00:00:42,363][14652] Decorrelating experience for 64 frames...
+[2025-10-19 00:00:42,656][14652] Decorrelating experience for 96 frames...
+[2025-10-19 00:00:46,070][14521] Fps is (10 sec: 0.0, 60 sec: 0.0, 300 sec: 0.0). Total num frames: 0. Throughput: 0: 1.2. Samples: 6. Policy #0 lag: (min: -1.0, avg: -1.0, max: -1.0)
+[2025-10-19 00:00:46,074][14521] Avg episode reward: [(0, '-0.077')]
+[2025-10-19 00:00:47,558][14521] Heartbeat connected on Batcher_0
+[2025-10-19 00:00:47,574][14521] Heartbeat connected on RolloutWorker_w0
+[2025-10-19 00:00:47,576][14521] Heartbeat connected on InferenceWorker_p0-w0
+[2025-10-19 00:00:49,568][14634] Signal inference workers to stop experience collection...
+[2025-10-19 00:00:49,583][14651] InferenceWorker_p0-w0: stopping experience collection
+[2025-10-19 00:00:49,772][14634] Signal inference workers to resume experience collection...
+[2025-10-19 00:00:49,775][14651] InferenceWorker_p0-w0: resuming experience collection
+[2025-10-19 00:00:50,283][14521] Heartbeat connected on LearnerWorker_p0
+[2025-10-19 00:00:51,070][14521] Fps is (10 sec: 819.2, 60 sec: 819.2, 300 sec: 819.2). Total num frames: 8192. Throughput: 0: 209.4. Samples: 2094. Policy #0 lag: (min: 0.0, avg: 0.0, max: 0.0)
+[2025-10-19 00:00:51,072][14521] Avg episode reward: [(0, '-0.175')]
+[2025-10-19 00:00:55,025][14634] Stopping Batcher_0...
+[2025-10-19 00:00:55,028][14634] Saving train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000004_16384.pth...
+[2025-10-19 00:00:55,025][14521] Component Batcher_0 stopped!
+[2025-10-19 00:00:55,028][14634] Loop batcher_evt_loop terminating...
+[2025-10-19 00:00:55,058][14651] Weights refcount: 2 0
+[2025-10-19 00:00:55,060][14651] Stopping InferenceWorker_p0-w0...
+[2025-10-19 00:00:55,060][14521] Component InferenceWorker_p0-w0 stopped!
+[2025-10-19 00:00:55,064][14651] Loop inference_proc0-0_evt_loop terminating...
+[2025-10-19 00:00:55,217][14634] Removing train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000004_16384.pth
+[2025-10-19 00:00:55,236][14634] Saving train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000004_16384.pth...
+[2025-10-19 00:00:55,291][14652] Stopping RolloutWorker_w0...
+[2025-10-19 00:00:55,292][14652] Loop rollout_proc0_evt_loop terminating...
+[2025-10-19 00:00:55,293][14521] Component RolloutWorker_w0 stopped!
+[2025-10-19 00:00:55,408][14634] Removing train_dir/deadly_corridor_experiment/checkpoint_p0/checkpoint_000000004_16384.pth
+[2025-10-19 00:00:55,432][14634] Stopping LearnerWorker_p0...
+[2025-10-19 00:00:55,433][14521] Component LearnerWorker_p0 stopped!
+[2025-10-19 00:00:55,434][14634] Loop learner_proc0_evt_loop terminating...
+[2025-10-19 00:00:55,435][14521] Waiting for process learner_proc0 to stop...
+[2025-10-19 00:00:56,764][14521] Waiting for process inference_proc0-0 to join...
+[2025-10-19 00:00:56,766][14521] Waiting for process rollout_proc0 to join...
+[2025-10-19 00:00:56,768][14521] Batcher 0 profile tree view:
+batching: 0.0902, releasing_batches: 0.0034
+[2025-10-19 00:00:56,770][14521] InferenceWorker_p0-w0 profile tree view:
+wait_policy: 0.0000
+  wait_policy_total: 2.6831
+update_model: 0.2012
+  weight_update: 0.0345
+one_step: 0.0035
+  handle_policy_step: 10.5613
+    deserialize: 0.1239, stack: 0.0335, obs_to_device_normalize: 1.8416, forward: 7.2182, send_messages: 0.3024
+    prepare_outputs: 0.8086
+      to_cpu: 0.5498
+[2025-10-19 00:00:56,772][14521] Learner 0 profile tree view:
+misc: 0.0000, prepare_batch: 1.6133
+train: 2.9548
+  epoch_init: 0.0000, minibatch_init: 0.0000, losses_postprocess: 0.0028, kl_divergence: 0.0235, after_optimizer: 0.1886
+  calculate_losses: 0.7724
+    losses_init: 0.0000, forward_head: 0.4685, bptt_initial: 0.1183, tail: 0.0694, advantages_returns: 0.0015, losses: 0.0988
+    bptt: 0.0152
+      bptt_forward_core: 0.0148
+  update: 1.9656
+    clip: 0.1359
+[2025-10-19 00:00:56,773][14521] RolloutWorker_w0 profile tree view:
+wait_for_trajectories: 0.0049, enqueue_policy_requests: 0.4166, env_step: 6.8453, overhead: 0.2312, complete_rollouts: 0.0163
+save_policy_outputs: 0.2837
+  split_output_tensors: 0.1146
+[2025-10-19 00:00:56,775][14521] Loop Runner_EvtLoop terminating...
+[2025-10-19 00:00:56,776][14521] Runner profile tree view:
+main_loop: 29.2021
+[2025-10-19 00:00:56,777][14521] Collected {0: 16384}, FPS: 561.1