Upload mouse-run-run-2-ssm: trained agents, rollouts, evaluation, cross-architecture study
a1f2a94 verified | {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0000/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0000/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0000/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0000.attempt_01", "seed": 0, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.00037524006884837036, "chaser_grad_norm": 0.15864729136228561, "chaser_new_fields": 86.55000305175781, "chaser_partner_vision": 0.0, "chaser_return": 7.260000705718994, "chaser_subspace_norm": 222.2711639404297, "collisions_per_episode": 1.274999976158142, "entropy": 0.007616071961820126, "explorer_grad_norm": 3.455117702484131, "explorer_new_fields": 93.32500457763672, "explorer_partner_vision": 0.0, "explorer_return": 89.5374984741211, "explorer_subspace_norm": 281.7393798828125, "final_distance": 4.3824381828308105, "max_param_abs": 1.7958338260650635, "policy_loss": -0.000521259798574647, "value_loss": 0.17206959798932076}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:41.687406+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 5.087962627410889, "chaser_new_fields": 85.79999542236328, "chaser_partner_vision": 0.0, "chaser_return": 6.949995517730713, "chaser_stuck_run_steps": 0.0, "collisions_per_episode": 3.3399999141693115, "degenerate_episodes": 85, "degenerate_fraction": 0.85, "episodes": 100, "explorer_new_fields": 32.97999954223633, "explorer_partner_vision": 0.0, "explorer_return": -1.25, "explorer_stuck_run_steps": 2.7300000190734863, "final_distance": 5.153793811798096, "same_state_run_steps": 0.0}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.non_social.seed_0000.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0000"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0000/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0000/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0000/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0000.attempt_01", "seed": 0, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.00037524006884837036, "chaser_grad_norm": 0.15864729136228561, "chaser_new_fields": 86.55000305175781, "chaser_partner_vision": 0.0, "chaser_return": 7.260000705718994, "chaser_subspace_norm": 222.2711639404297, "collisions_per_episode": 1.274999976158142, "entropy": 0.007616071961820126, "explorer_grad_norm": 3.455117702484131, "explorer_new_fields": 93.32500457763672, "explorer_partner_vision": 0.0, "explorer_return": 89.5374984741211, "explorer_subspace_norm": 281.7393798828125, "final_distance": 4.3824381828308105, "max_param_abs": 1.7958338260650635, "policy_loss": -0.000521259798574647, "value_loss": 0.17206959798932076}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:41.794059+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.045877933502197, "chaser_new_fields": 33.65999984741211, "chaser_partner_vision": 0.0, "chaser_return": -3.3079988956451416, "chaser_stuck_run_steps": 2.569999933242798, "collisions_per_episode": 2.9600000381469727, "degenerate_episodes": 88, "degenerate_fraction": 0.88, "episodes": 100, "explorer_new_fields": 92.25, "explorer_partner_vision": 0.0, "explorer_return": 87.4000015258789, "explorer_stuck_run_steps": 0.0, "final_distance": 4.336213111877441, "same_state_run_steps": 0.0}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.non_social.seed_0000.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0000"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0001/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0001/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0001/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0001.attempt_01", "seed": 1, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.000269684101368739, "chaser_grad_norm": 0.21666903048753738, "chaser_new_fields": 65.0999984741211, "chaser_partner_vision": 0.0, "chaser_return": 3.000000238418579, "chaser_subspace_norm": 145.25376892089844, "collisions_per_episode": 0.42500001192092896, "entropy": 0.006687503308057785, "explorer_grad_norm": 2.867681562900543, "explorer_new_fields": 90.07500457763672, "explorer_partner_vision": 0.0, "explorer_return": 84.7750015258789, "explorer_subspace_norm": 241.7664337158203, "final_distance": 4.018166542053223, "max_param_abs": 1.983300805091858, "policy_loss": 3.472161620265979e-05, "value_loss": 0.06099161971360445}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:41.911530+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 4.789496898651123, "chaser_new_fields": 64.63999938964844, "chaser_partner_vision": 0.0, "chaser_return": 2.80199933052063, "chaser_stuck_run_steps": 0.07999999821186066, "collisions_per_episode": 2.549999952316284, "degenerate_episodes": 87, "degenerate_fraction": 0.87, "episodes": 100, "explorer_new_fields": 32.90999984741211, "explorer_partner_vision": 0.0, "explorer_return": -1.25, "explorer_stuck_run_steps": 2.690000057220459, "final_distance": 4.906148433685303, "same_state_run_steps": 0.009999999776482582}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.non_social.seed_0001.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0001"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0001/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0001/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0001/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0001.attempt_01", "seed": 1, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.000269684101368739, "chaser_grad_norm": 0.21666903048753738, "chaser_new_fields": 65.0999984741211, "chaser_partner_vision": 0.0, "chaser_return": 3.000000238418579, "chaser_subspace_norm": 145.25376892089844, "collisions_per_episode": 0.42500001192092896, "entropy": 0.006687503308057785, "explorer_grad_norm": 2.867681562900543, "explorer_new_fields": 90.07500457763672, "explorer_partner_vision": 0.0, "explorer_return": 84.7750015258789, "explorer_subspace_norm": 241.7664337158203, "final_distance": 4.018166542053223, "max_param_abs": 1.983300805091858, "policy_loss": 3.472161620265979e-05, "value_loss": 0.06099161971360445}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:42.017113+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.125200271606445, "chaser_new_fields": 34.04999923706055, "chaser_partner_vision": 0.0, "chaser_return": -3.231998920440674, "chaser_stuck_run_steps": 2.6599998474121094, "collisions_per_episode": 2.559999942779541, "degenerate_episodes": 89, "degenerate_fraction": 0.89, "episodes": 100, "explorer_new_fields": 88.63999938964844, "explorer_partner_vision": 0.0, "explorer_return": 81.95500183105469, "explorer_stuck_run_steps": 0.0, "final_distance": 4.807648658752441, "same_state_run_steps": 0.0}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.non_social.seed_0001.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0001"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0002/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0002/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0002/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0002.attempt_01", "seed": 2, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.000490981027894577, "chaser_grad_norm": 0.05319501273334026, "chaser_new_fields": 93.8499984741211, "chaser_partner_vision": 0.0, "chaser_return": 8.71500015258789, "chaser_subspace_norm": 113.91453552246094, "collisions_per_episode": 0.800000011920929, "entropy": 0.0015642183716408908, "explorer_grad_norm": 3.073855459690094, "explorer_new_fields": 90.5250015258789, "explorer_partner_vision": 0.0, "explorer_return": 85.45000457763672, "explorer_subspace_norm": 177.7625732421875, "final_distance": 5.066453456878662, "max_param_abs": 1.5996028184890747, "policy_loss": -0.00041541958586321925, "value_loss": 0.08431601151823997}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:42.129397+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 5.187206268310547, "chaser_new_fields": 92.52999877929688, "chaser_partner_vision": 0.0, "chaser_return": 8.309996604919434, "chaser_stuck_run_steps": 0.0, "collisions_per_episode": 2.950000047683716, "degenerate_episodes": 83, "degenerate_fraction": 0.83, "episodes": 100, "explorer_new_fields": 33.02000045776367, "explorer_partner_vision": 0.0, "explorer_return": -0.9950000047683716, "explorer_stuck_run_steps": 2.679999828338623, "final_distance": 5.139909744262695, "same_state_run_steps": 0.0}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.non_social.seed_0002.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0002"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0002/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0002/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0002/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0002.attempt_01", "seed": 2, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.000490981027894577, "chaser_grad_norm": 0.05319501273334026, "chaser_new_fields": 93.8499984741211, "chaser_partner_vision": 0.0, "chaser_return": 8.71500015258789, "chaser_subspace_norm": 113.91453552246094, "collisions_per_episode": 0.800000011920929, "entropy": 0.0015642183716408908, "explorer_grad_norm": 3.073855459690094, "explorer_new_fields": 90.5250015258789, "explorer_partner_vision": 0.0, "explorer_return": 85.45000457763672, "explorer_subspace_norm": 177.7625732421875, "final_distance": 5.066453456878662, "max_param_abs": 1.5996028184890747, "policy_loss": -0.00041541958586321925, "value_loss": 0.08431601151823997}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:42.234689+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.042456150054932, "chaser_new_fields": 33.75, "chaser_partner_vision": 0.0, "chaser_return": -3.3079988956451416, "chaser_stuck_run_steps": 2.6499998569488525, "collisions_per_episode": 2.799999952316284, "degenerate_episodes": 89, "degenerate_fraction": 0.89, "episodes": 100, "explorer_new_fields": 88.62999725341797, "explorer_partner_vision": 0.0, "explorer_return": 81.92499542236328, "explorer_stuck_run_steps": 0.0, "final_distance": 4.980891227722168, "same_state_run_steps": 0.0}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.non_social.seed_0002.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0002"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0003/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0003/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0003/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0003.attempt_01", "seed": 3, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 5.2467806275657836e-05, "chaser_grad_norm": 0.2821537107229233, "chaser_new_fields": 94.625, "chaser_partner_vision": 0.0, "chaser_return": 8.90000057220459, "chaser_subspace_norm": 127.75957489013672, "collisions_per_episode": 0.42500001192092896, "entropy": 0.0007984262483660132, "explorer_grad_norm": 2.997040629386902, "explorer_new_fields": 92.32500457763672, "explorer_partner_vision": 0.0, "explorer_return": 88.2249984741211, "explorer_subspace_norm": 227.22293090820312, "final_distance": 5.372359752655029, "max_param_abs": 1.7984249591827393, "policy_loss": -5.933013869707793e-05, "value_loss": 0.06194356083869934}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:42.346673+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 5.163653373718262, "chaser_new_fields": 91.86000061035156, "chaser_partner_vision": 0.0, "chaser_return": 8.137996673583984, "chaser_stuck_run_steps": 0.0, "collisions_per_episode": 3.379999876022339, "degenerate_episodes": 88, "degenerate_fraction": 0.88, "episodes": 100, "explorer_new_fields": 33.15999984741211, "explorer_partner_vision": 0.0, "explorer_return": -1.159999966621399, "explorer_stuck_run_steps": 2.7200000286102295, "final_distance": 4.926669597625732, "same_state_run_steps": 0.0}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.non_social.seed_0003.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0003"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0003/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0003/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0003/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0003.attempt_01", "seed": 3, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 5.2467806275657836e-05, "chaser_grad_norm": 0.2821537107229233, "chaser_new_fields": 94.625, "chaser_partner_vision": 0.0, "chaser_return": 8.90000057220459, "chaser_subspace_norm": 127.75957489013672, "collisions_per_episode": 0.42500001192092896, "entropy": 0.0007984262483660132, "explorer_grad_norm": 2.997040629386902, "explorer_new_fields": 92.32500457763672, "explorer_partner_vision": 0.0, "explorer_return": 88.2249984741211, "explorer_subspace_norm": 227.22293090820312, "final_distance": 5.372359752655029, "max_param_abs": 1.7984249591827393, "policy_loss": -5.933013869707793e-05, "value_loss": 0.06194356083869934}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:42.451820+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.149752616882324, "chaser_new_fields": 33.869998931884766, "chaser_partner_vision": 0.0, "chaser_return": -3.2639989852905273, "chaser_stuck_run_steps": 2.609999895095825, "collisions_per_episode": 2.2200000286102295, "degenerate_episodes": 86, "degenerate_fraction": 0.86, "episodes": 100, "explorer_new_fields": 91.75, "explorer_partner_vision": 0.0, "explorer_return": 86.86000061035156, "explorer_stuck_run_steps": 0.0, "final_distance": 5.063153266906738, "same_state_run_steps": 0.0}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.non_social.seed_0003.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0003"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0004/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0004/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0004/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0004.attempt_01", "seed": 4, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.0002866870536522015, "chaser_grad_norm": 1.2475789934396744, "chaser_new_fields": 89.9000015258789, "chaser_partner_vision": 0.0, "chaser_return": 7.9200005531311035, "chaser_subspace_norm": 131.41030883789062, "collisions_per_episode": 1.25, "entropy": 0.002184464654419571, "explorer_grad_norm": 3.1244588792324066, "explorer_new_fields": 88.45000457763672, "explorer_partner_vision": 0.0, "explorer_return": 82.2249984741211, "explorer_subspace_norm": 223.06732177734375, "final_distance": 5.393571376800537, "max_param_abs": 1.7349889278411865, "policy_loss": -0.00031348039517808957, "value_loss": 0.32997676730155945}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:42.564447+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 5.139449119567871, "chaser_new_fields": 88.6500015258789, "chaser_partner_vision": 0.0, "chaser_return": 7.539994716644287, "chaser_stuck_run_steps": 0.0, "collisions_per_episode": 3.009999990463257, "degenerate_episodes": 84, "degenerate_fraction": 0.84, "episodes": 100, "explorer_new_fields": 32.82999801635742, "explorer_partner_vision": 0.0, "explorer_return": -1.4299999475479126, "explorer_stuck_run_steps": 2.75, "final_distance": 4.528690814971924, "same_state_run_steps": 0.0}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.non_social.seed_0004.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0004"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0004/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0004/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0004/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0004.attempt_01", "seed": 4, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.0002866870536522015, "chaser_grad_norm": 1.2475789934396744, "chaser_new_fields": 89.9000015258789, "chaser_partner_vision": 0.0, "chaser_return": 7.9200005531311035, "chaser_subspace_norm": 131.41030883789062, "collisions_per_episode": 1.25, "entropy": 0.002184464654419571, "explorer_grad_norm": 3.1244588792324066, "explorer_new_fields": 88.45000457763672, "explorer_partner_vision": 0.0, "explorer_return": 82.2249984741211, "explorer_subspace_norm": 223.06732177734375, "final_distance": 5.393571376800537, "max_param_abs": 1.7349889278411865, "policy_loss": -0.00031348039517808957, "value_loss": 0.32997676730155945}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:42.670069+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 4.992640972137451, "chaser_new_fields": 33.619998931884766, "chaser_partner_vision": 0.0, "chaser_return": -3.3219988346099854, "chaser_stuck_run_steps": 2.6599998474121094, "collisions_per_episode": 2.559999942779541, "degenerate_episodes": 90, "degenerate_fraction": 0.9, "episodes": 100, "explorer_new_fields": 87.11000061035156, "explorer_partner_vision": 0.0, "explorer_return": 79.95999908447266, "explorer_stuck_run_steps": 0.009999999776482582, "final_distance": 4.85975456237793, "same_state_run_steps": 0.0}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.non_social.seed_0004.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0004"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0005/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0005/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0005/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0005.attempt_01", "seed": 5, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.000728657579465164, "chaser_grad_norm": 0.17576856166124344, "chaser_new_fields": 91.55000305175781, "chaser_partner_vision": 0.0, "chaser_return": 8.27500057220459, "chaser_subspace_norm": 147.91407775878906, "collisions_per_episode": 0.42500001192092896, "entropy": 0.00442305754404515, "explorer_grad_norm": 3.893365442752838, "explorer_new_fields": 92.82500457763672, "explorer_partner_vision": 0.0, "explorer_return": 88.9375, "explorer_subspace_norm": 290.2926025390625, "final_distance": 3.6594765186309814, "max_param_abs": 1.9001684188842773, "policy_loss": -0.0015874679654643842, "value_loss": 0.08237634599208832}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:42.781872+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 5.119772911071777, "chaser_new_fields": 89.22000122070312, "chaser_partner_vision": 0.0, "chaser_return": 7.621995449066162, "chaser_stuck_run_steps": 0.0, "collisions_per_episode": 3.2100000381469727, "degenerate_episodes": 86, "degenerate_fraction": 0.86, "episodes": 100, "explorer_new_fields": 33.22999954223633, "explorer_partner_vision": 0.0, "explorer_return": -0.9049999713897705, "explorer_stuck_run_steps": 2.700000047683716, "final_distance": 4.226670742034912, "same_state_run_steps": 0.0}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.non_social.seed_0005.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0005"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0005/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0005/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0005/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0005.attempt_01", "seed": 5, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.000728657579465164, "chaser_grad_norm": 0.17576856166124344, "chaser_new_fields": 91.55000305175781, "chaser_partner_vision": 0.0, "chaser_return": 8.27500057220459, "chaser_subspace_norm": 147.91407775878906, "collisions_per_episode": 0.42500001192092896, "entropy": 0.00442305754404515, "explorer_grad_norm": 3.893365442752838, "explorer_new_fields": 92.82500457763672, "explorer_partner_vision": 0.0, "explorer_return": 88.9375, "explorer_subspace_norm": 290.2926025390625, "final_distance": 3.6594765186309814, "max_param_abs": 1.9001684188842773, "policy_loss": -0.0015874679654643842, "value_loss": 0.08237634599208832}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:42.887127+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.100900173187256, "chaser_new_fields": 34.029998779296875, "chaser_partner_vision": 0.0, "chaser_return": -3.2419989109039307, "chaser_stuck_run_steps": 2.700000047683716, "collisions_per_episode": 2.5799999237060547, "degenerate_episodes": 90, "degenerate_fraction": 0.9, "episodes": 100, "explorer_new_fields": 90.66999816894531, "explorer_partner_vision": 0.0, "explorer_return": 85.20999908447266, "explorer_stuck_run_steps": 0.009999999776482582, "final_distance": 4.707333564758301, "same_state_run_steps": 0.0}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.non_social.seed_0005.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0005"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0006/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0006/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0006/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0006.attempt_01", "seed": 6, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.0003681715173087241, "chaser_grad_norm": 0.07220392860472202, "chaser_new_fields": 85.4749984741211, "chaser_partner_vision": 0.0, "chaser_return": 7.070001125335693, "chaser_subspace_norm": 133.82351684570312, "collisions_per_episode": 0.574999988079071, "entropy": 0.003955341002438217, "explorer_grad_norm": 3.9367371797561646, "explorer_new_fields": 95.57500457763672, "explorer_partner_vision": 0.0, "explorer_return": 92.95000457763672, "explorer_subspace_norm": 349.3121032714844, "final_distance": 4.353031158447266, "max_param_abs": 1.8634331226348877, "policy_loss": -0.0003699150271359031, "value_loss": 0.06900897808372974}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:42.998955+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 5.003993034362793, "chaser_new_fields": 84.61000061035156, "chaser_partner_vision": 0.0, "chaser_return": 6.747995853424072, "chaser_stuck_run_steps": 0.0, "collisions_per_episode": 3.0299999713897705, "degenerate_episodes": 82, "degenerate_fraction": 0.82, "episodes": 100, "explorer_new_fields": 33.119998931884766, "explorer_partner_vision": 0.0, "explorer_return": -1.1749999523162842, "explorer_stuck_run_steps": 2.700000047683716, "final_distance": 4.522829055786133, "same_state_run_steps": 0.0}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.non_social.seed_0006.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0006"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0006/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0006/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0006/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0006.attempt_01", "seed": 6, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.0003681715173087241, "chaser_grad_norm": 0.07220392860472202, "chaser_new_fields": 85.4749984741211, "chaser_partner_vision": 0.0, "chaser_return": 7.070001125335693, "chaser_subspace_norm": 133.82351684570312, "collisions_per_episode": 0.574999988079071, "entropy": 0.003955341002438217, "explorer_grad_norm": 3.9367371797561646, "explorer_new_fields": 95.57500457763672, "explorer_partner_vision": 0.0, "explorer_return": 92.95000457763672, "explorer_subspace_norm": 349.3121032714844, "final_distance": 4.353031158447266, "max_param_abs": 1.8634331226348877, "policy_loss": -0.0003699150271359031, "value_loss": 0.06900897808372974}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:43.103851+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.113889217376709, "chaser_new_fields": 33.57999801635742, "chaser_partner_vision": 0.0, "chaser_return": -3.339998722076416, "chaser_stuck_run_steps": 2.6399998664855957, "collisions_per_episode": 2.759999990463257, "degenerate_episodes": 87, "degenerate_fraction": 0.87, "episodes": 100, "explorer_new_fields": 93.8499984741211, "explorer_partner_vision": 0.0, "explorer_return": 89.7699966430664, "explorer_stuck_run_steps": 0.0, "final_distance": 4.519081115722656, "same_state_run_steps": 0.0}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.non_social.seed_0006.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0006"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0007/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0007/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0007/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0007.attempt_01", "seed": 7, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.00011773648748203414, "chaser_grad_norm": 0.187929792329669, "chaser_new_fields": 90.67500305175781, "chaser_partner_vision": 0.0, "chaser_return": 8.090001106262207, "chaser_subspace_norm": 194.9869842529297, "collisions_per_episode": 0.675000011920929, "entropy": 0.005327099352143705, "explorer_grad_norm": 3.9027020931243896, "explorer_new_fields": 90.80000305175781, "explorer_partner_vision": 0.0, "explorer_return": 85.86250305175781, "explorer_subspace_norm": 212.644287109375, "final_distance": 5.175065994262695, "max_param_abs": 1.5573863983154297, "policy_loss": -0.00039410117392435495, "value_loss": 0.0847784411162138}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:43.215842+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 5.164993762969971, "chaser_new_fields": 88.56999969482422, "chaser_partner_vision": 0.0, "chaser_return": 7.527995586395264, "chaser_stuck_run_steps": 0.0, "collisions_per_episode": 3.119999885559082, "degenerate_episodes": 84, "degenerate_fraction": 0.84, "episodes": 100, "explorer_new_fields": 32.70000076293945, "explorer_partner_vision": 0.0, "explorer_return": -1.7599999904632568, "explorer_stuck_run_steps": 2.759999990463257, "final_distance": 4.968615531921387, "same_state_run_steps": 0.0}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.non_social.seed_0007.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0007"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0007/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0007/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0007/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0007.attempt_01", "seed": 7, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.00011773648748203414, "chaser_grad_norm": 0.187929792329669, "chaser_new_fields": 90.67500305175781, "chaser_partner_vision": 0.0, "chaser_return": 8.090001106262207, "chaser_subspace_norm": 194.9869842529297, "collisions_per_episode": 0.675000011920929, "entropy": 0.005327099352143705, "explorer_grad_norm": 3.9027020931243896, "explorer_new_fields": 90.80000305175781, "explorer_partner_vision": 0.0, "explorer_return": 85.86250305175781, "explorer_subspace_norm": 212.644287109375, "final_distance": 5.175065994262695, "max_param_abs": 1.5573863983154297, "policy_loss": -0.00039410117392435495, "value_loss": 0.0847784411162138}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:43.321084+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.073260307312012, "chaser_new_fields": 34.119998931884766, "chaser_partner_vision": 0.0, "chaser_return": -3.21799898147583, "chaser_stuck_run_steps": 2.669999837875366, "collisions_per_episode": 2.5399999618530273, "degenerate_episodes": 88, "degenerate_fraction": 0.88, "episodes": 100, "explorer_new_fields": 88.0199966430664, "explorer_partner_vision": 0.0, "explorer_return": 81.1449966430664, "explorer_stuck_run_steps": 0.0, "final_distance": 4.29344367980957, "same_state_run_steps": 0.0}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.non_social.seed_0007.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0007"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0008/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0008/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0008/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0008.attempt_01", "seed": 8, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.000258021026315447, "chaser_grad_norm": 0.09883276652544737, "chaser_new_fields": 89.5999984741211, "chaser_partner_vision": 0.0, "chaser_return": 7.885000705718994, "chaser_subspace_norm": 170.72024536132812, "collisions_per_episode": 0.550000011920929, "entropy": 0.002424428064841777, "explorer_grad_norm": 2.3192933201789856, "explorer_new_fields": 91.2249984741211, "explorer_partner_vision": 0.0, "explorer_return": 86.57500457763672, "explorer_subspace_norm": 261.4172668457031, "final_distance": 5.970248699188232, "max_param_abs": 1.7791407108306885, "policy_loss": -4.3978677695122315e-06, "value_loss": 0.05447797849774361}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:43.432932+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 5.1030426025390625, "chaser_new_fields": 87.1500015258789, "chaser_partner_vision": 0.0, "chaser_return": 7.241995811462402, "chaser_stuck_run_steps": 0.0, "collisions_per_episode": 3.009999990463257, "degenerate_episodes": 85, "degenerate_fraction": 0.85, "episodes": 100, "explorer_new_fields": 33.25, "explorer_partner_vision": 0.0, "explorer_return": -0.7099999785423279, "explorer_stuck_run_steps": 2.619999885559082, "final_distance": 5.472009181976318, "same_state_run_steps": 0.0}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.non_social.seed_0008.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0008"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0008/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0008/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0008/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0008.attempt_01", "seed": 8, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.000258021026315447, "chaser_grad_norm": 0.09883276652544737, "chaser_new_fields": 89.5999984741211, "chaser_partner_vision": 0.0, "chaser_return": 7.885000705718994, "chaser_subspace_norm": 170.72024536132812, "collisions_per_episode": 0.550000011920929, "entropy": 0.002424428064841777, "explorer_grad_norm": 2.3192933201789856, "explorer_new_fields": 91.2249984741211, "explorer_partner_vision": 0.0, "explorer_return": 86.57500457763672, "explorer_subspace_norm": 261.4172668457031, "final_distance": 5.970248699188232, "max_param_abs": 1.7791407108306885, "policy_loss": -4.3978677695122315e-06, "value_loss": 0.05447797849774361}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:43.538537+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.0984063148498535, "chaser_new_fields": 33.96999740600586, "chaser_partner_vision": 0.0, "chaser_return": -3.2679989337921143, "chaser_stuck_run_steps": 2.7100000381469727, "collisions_per_episode": 2.6599998474121094, "degenerate_episodes": 86, "degenerate_fraction": 0.86, "episodes": 100, "explorer_new_fields": 89.54000091552734, "explorer_partner_vision": 0.0, "explorer_return": 83.31999969482422, "explorer_stuck_run_steps": 0.03999999910593033, "final_distance": 4.667797088623047, "same_state_run_steps": 0.009999999776482582}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.non_social.seed_0008.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0008"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0009/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0009/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0009/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0009.attempt_01", "seed": 9, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 5.256045653112551e-05, "chaser_grad_norm": 0.14516921155154705, "chaser_new_fields": 85.32500457763672, "chaser_partner_vision": 0.0, "chaser_return": 7.03000020980835, "chaser_subspace_norm": 131.6070098876953, "collisions_per_episode": 1.7000000476837158, "entropy": 0.0016931382124312222, "explorer_grad_norm": 3.502101480960846, "explorer_new_fields": 89.6500015258789, "explorer_partner_vision": 0.0, "explorer_return": 83.875, "explorer_subspace_norm": 238.18890380859375, "final_distance": 4.5569353103637695, "max_param_abs": 1.730488896369934, "policy_loss": -3.0939103524474376e-05, "value_loss": 0.1814001314342022}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:43.650356+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 5.059107780456543, "chaser_new_fields": 83.88999938964844, "chaser_partner_vision": 0.0, "chaser_return": 6.535995960235596, "chaser_stuck_run_steps": 0.0, "collisions_per_episode": 3.93999981880188, "degenerate_episodes": 85, "degenerate_fraction": 0.85, "episodes": 100, "explorer_new_fields": 32.72999954223633, "explorer_partner_vision": 0.0, "explorer_return": -1.7450000047683716, "explorer_stuck_run_steps": 2.75, "final_distance": 4.74109411239624, "same_state_run_steps": 0.0}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.non_social.seed_0009.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0009"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0009/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/non_social/seed_0009/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "none", "task": "non_social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/non_social/seed_0009/attempt_01", "run_id": "mouse-run-run-2-ssm.non_social.seed_0009.attempt_01", "seed": 9, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 5.256045653112551e-05, "chaser_grad_norm": 0.14516921155154705, "chaser_new_fields": 85.32500457763672, "chaser_partner_vision": 0.0, "chaser_return": 7.03000020980835, "chaser_subspace_norm": 131.6070098876953, "collisions_per_episode": 1.7000000476837158, "entropy": 0.0016931382124312222, "explorer_grad_norm": 3.502101480960846, "explorer_new_fields": 89.6500015258789, "explorer_partner_vision": 0.0, "explorer_return": 83.875, "explorer_subspace_norm": 238.18890380859375, "final_distance": 4.5569353103637695, "max_param_abs": 1.730488896369934, "policy_loss": -3.0939103524474376e-05, "value_loss": 0.1814001314342022}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:43.755466+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 4.984620094299316, "chaser_new_fields": 33.869998931884766, "chaser_partner_vision": 0.0, "chaser_return": -3.269998788833618, "chaser_stuck_run_steps": 2.549999952316284, "collisions_per_episode": 3.069999933242798, "degenerate_episodes": 84, "degenerate_fraction": 0.84, "episodes": 100, "explorer_new_fields": 88.47000122070312, "explorer_partner_vision": 0.0, "explorer_return": 81.68499755859375, "explorer_stuck_run_steps": 0.009999999776482582, "final_distance": 4.66384744644165, "same_state_run_steps": 0.0}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.non_social.seed_0009.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "non_social", "unit_id": "non_social/seed_0009"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0000/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0000/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0000/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0000.attempt_01", "seed": 0, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.003996992898350982, "chaser_grad_norm": 0.4637064263224602, "chaser_new_fields": 51.17499923706055, "chaser_partner_vision": 0.9815000295639038, "chaser_return": 29.962499618530273, "chaser_subspace_norm": 167.948974609375, "collisions_per_episode": 27.625, "entropy": 0.19277657940983772, "explorer_grad_norm": 1.111926943063736, "explorer_new_fields": 55.95000076293945, "explorer_partner_vision": 0.9815000295639038, "explorer_return": 16.024999618530273, "explorer_subspace_norm": 77.47383880615234, "final_distance": 1.6499801874160767, "max_param_abs": 1.5146597623825073, "policy_loss": -0.0042748258374558645, "value_loss": 4.675158381462097}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:43.875664+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 1.742310643196106, "chaser_new_fields": 25.059999465942383, "chaser_partner_vision": 0.9449999928474426, "chaser_return": 36.04600143432617, "chaser_stuck_run_steps": 1.5499999523162842, "collisions_per_episode": 37.81999969482422, "degenerate_episodes": 99, "degenerate_fraction": 0.99, "episodes": 100, "explorer_new_fields": 22.19999885559082, "explorer_partner_vision": 0.9449999928474426, "explorer_return": -41.21999740600586, "explorer_stuck_run_steps": 3.7200000286102295, "final_distance": 1.7924898862838745, "same_state_run_steps": 0.17000000178813934}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.social.seed_0000.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0000"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0000/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0000/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0000/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0000.attempt_01", "seed": 0, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.003996992898350982, "chaser_grad_norm": 0.4637064263224602, "chaser_new_fields": 51.17499923706055, "chaser_partner_vision": 0.9815000295639038, "chaser_return": 29.962499618530273, "chaser_subspace_norm": 167.948974609375, "collisions_per_episode": 27.625, "entropy": 0.19277657940983772, "explorer_grad_norm": 1.111926943063736, "explorer_new_fields": 55.95000076293945, "explorer_partner_vision": 0.9815000295639038, "explorer_return": 16.024999618530273, "explorer_subspace_norm": 77.47383880615234, "final_distance": 1.6499801874160767, "max_param_abs": 1.5146597623825073, "policy_loss": -0.0042748258374558645, "value_loss": 4.675158381462097}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:43.989227+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.45292854309082, "chaser_new_fields": 33.869998931884766, "chaser_partner_vision": 0.2606000006198883, "chaser_return": -2.665999174118042, "chaser_stuck_run_steps": 2.6499998569488525, "collisions_per_episode": 0.5199999809265137, "degenerate_episodes": 88, "degenerate_fraction": 0.88, "episodes": 100, "explorer_new_fields": 61.90999984741211, "explorer_partner_vision": 0.2606000006198883, "explorer_return": 42.36499786376953, "explorer_stuck_run_steps": 0.20999999344348907, "final_distance": 5.397023677825928, "same_state_run_steps": 0.009999999776482582}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.social.seed_0000.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0000"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0001/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0001/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0001/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0001.attempt_01", "seed": 1, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.005803498026352827, "chaser_grad_norm": 1.2199710309505463, "chaser_new_fields": 52.47500228881836, "chaser_partner_vision": 0.9690000414848328, "chaser_return": 18.967500686645508, "chaser_subspace_norm": 212.3469696044922, "collisions_per_episode": 17.325000762939453, "entropy": 0.12894443795084953, "explorer_grad_norm": 1.7083011865615845, "explorer_new_fields": 64.0999984741211, "explorer_partner_vision": 0.9690000414848328, "explorer_return": 32.95000076293945, "explorer_subspace_norm": 74.30672454833984, "final_distance": 1.6290931701660156, "max_param_abs": 1.8514448404312134, "policy_loss": -0.002297161597198283, "value_loss": 3.5172311067581177}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:44.109239+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 2.132615566253662, "chaser_new_fields": 26.799999237060547, "chaser_partner_vision": 0.8789999485015869, "chaser_return": 30.349998474121094, "chaser_stuck_run_steps": 0.25, "collisions_per_episode": 32.31999969482422, "degenerate_episodes": 95, "degenerate_fraction": 0.95, "episodes": 100, "explorer_new_fields": 25.0, "explorer_partner_vision": 0.8789999485015869, "explorer_return": -34.52499771118164, "explorer_stuck_run_steps": 3.379999876022339, "final_distance": 2.1192781925201416, "same_state_run_steps": 0.07999999821186066}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.social.seed_0001.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0001"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0001/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0001/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0001/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0001.attempt_01", "seed": 1, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.005803498026352827, "chaser_grad_norm": 1.2199710309505463, "chaser_new_fields": 52.47500228881836, "chaser_partner_vision": 0.9690000414848328, "chaser_return": 18.967500686645508, "chaser_subspace_norm": 212.3469696044922, "collisions_per_episode": 17.325000762939453, "entropy": 0.12894443795084953, "explorer_grad_norm": 1.7083011865615845, "explorer_new_fields": 64.0999984741211, "explorer_partner_vision": 0.9690000414848328, "explorer_return": 32.95000076293945, "explorer_subspace_norm": 74.30672454833984, "final_distance": 1.6290931701660156, "max_param_abs": 1.8514448404312134, "policy_loss": -0.002297161597198283, "value_loss": 3.5172311067581177}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:44.222662+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.879758358001709, "chaser_new_fields": 33.86000061035156, "chaser_partner_vision": 0.2288999855518341, "chaser_return": -2.781999111175537, "chaser_stuck_run_steps": 2.629999876022339, "collisions_per_episode": 0.41999998688697815, "degenerate_episodes": 95, "degenerate_fraction": 0.95, "episodes": 100, "explorer_new_fields": 51.89999771118164, "explorer_partner_vision": 0.2288999855518341, "explorer_return": 27.399999618530273, "explorer_stuck_run_steps": 7.259999752044678, "final_distance": 6.539805889129639, "same_state_run_steps": 0.4099999964237213}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.social.seed_0001.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0001"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0002/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0002/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0002/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0002.attempt_01", "seed": 2, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.0023651940201148136, "chaser_grad_norm": 0.43530818074941635, "chaser_new_fields": 50.150001525878906, "chaser_partner_vision": 0.9787500500679016, "chaser_return": 15.404998779296875, "chaser_subspace_norm": 176.88015747070312, "collisions_per_episode": 14.40000057220459, "entropy": 0.1388651467859745, "explorer_grad_norm": 1.2929063141345978, "explorer_new_fields": 60.875, "explorer_partner_vision": 0.9787500500679016, "explorer_return": 29.575000762939453, "explorer_subspace_norm": 64.55570983886719, "final_distance": 1.5472310781478882, "max_param_abs": 1.5776029825210571, "policy_loss": -0.0026379052586094076, "value_loss": 3.196538031101227}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:44.342729+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 1.682454228401184, "chaser_new_fields": 24.850000381469727, "chaser_partner_vision": 0.9515999555587769, "chaser_return": 33.27199935913086, "chaser_stuck_run_steps": 1.090000033378601, "collisions_per_episode": 35.380001068115234, "degenerate_episodes": 99, "degenerate_fraction": 0.99, "episodes": 100, "explorer_new_fields": 22.439998626708984, "explorer_partner_vision": 0.9515999555587769, "explorer_return": -40.23999786376953, "explorer_stuck_run_steps": 3.6399998664855957, "final_distance": 1.4864531755447388, "same_state_run_steps": 0.5699999928474426}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.social.seed_0002.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0002"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0002/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0002/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0002/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0002.attempt_01", "seed": 2, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.0023651940201148136, "chaser_grad_norm": 0.43530818074941635, "chaser_new_fields": 50.150001525878906, "chaser_partner_vision": 0.9787500500679016, "chaser_return": 15.404998779296875, "chaser_subspace_norm": 176.88015747070312, "collisions_per_episode": 14.40000057220459, "entropy": 0.1388651467859745, "explorer_grad_norm": 1.2929063141345978, "explorer_new_fields": 60.875, "explorer_partner_vision": 0.9787500500679016, "explorer_return": 29.575000762939453, "explorer_subspace_norm": 64.55570983886719, "final_distance": 1.5472310781478882, "max_param_abs": 1.5776029825210571, "policy_loss": -0.0026379052586094076, "value_loss": 3.196538031101227}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:44.456355+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.790333271026611, "chaser_new_fields": 33.89999771118164, "chaser_partner_vision": 0.22279998660087585, "chaser_return": -2.717999219894409, "chaser_stuck_run_steps": 2.619999885559082, "collisions_per_episode": 0.47999998927116394, "degenerate_episodes": 91, "degenerate_fraction": 0.91, "episodes": 100, "explorer_new_fields": 57.32999801635742, "explorer_partner_vision": 0.22279998660087585, "explorer_return": 35.529998779296875, "explorer_stuck_run_steps": 0.5999999642372131, "final_distance": 6.158857345581055, "same_state_run_steps": 0.09999999403953552}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.social.seed_0002.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0002"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0003/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0003/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0003/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0003.attempt_01", "seed": 3, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.0021609386774343875, "chaser_grad_norm": 2.733183056116104, "chaser_new_fields": 57.5, "chaser_partner_vision": 0.9780000448226929, "chaser_return": 16.875, "chaser_subspace_norm": 173.6598663330078, "collisions_per_episode": 14.449999809265137, "entropy": 0.11295731365680695, "explorer_grad_norm": 1.9610493779182434, "explorer_new_fields": 65.95000457763672, "explorer_partner_vision": 0.9780000448226929, "explorer_return": 38.70000076293945, "explorer_subspace_norm": 79.3697738647461, "final_distance": 1.3376797437667847, "max_param_abs": 1.6915369033813477, "policy_loss": -0.0029535914400184993, "value_loss": 3.245455265045166}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:44.576283+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 1.7145581245422363, "chaser_new_fields": 27.899999618530273, "chaser_partner_vision": 0.9457000494003296, "chaser_return": 36.39099884033203, "chaser_stuck_run_steps": 2.4600000381469727, "collisions_per_episode": 37.709999084472656, "degenerate_episodes": 96, "degenerate_fraction": 0.96, "episodes": 100, "explorer_new_fields": 23.779998779296875, "explorer_partner_vision": 0.9457000494003296, "explorer_return": -39.82999801635742, "explorer_stuck_run_steps": 3.5299999713897705, "final_distance": 1.5691137313842773, "same_state_run_steps": 0.5}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.social.seed_0003.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0003"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0003/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0003/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0003/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0003.attempt_01", "seed": 3, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.0021609386774343875, "chaser_grad_norm": 2.733183056116104, "chaser_new_fields": 57.5, "chaser_partner_vision": 0.9780000448226929, "chaser_return": 16.875, "chaser_subspace_norm": 173.6598663330078, "collisions_per_episode": 14.449999809265137, "entropy": 0.11295731365680695, "explorer_grad_norm": 1.9610493779182434, "explorer_new_fields": 65.95000457763672, "explorer_partner_vision": 0.9780000448226929, "explorer_return": 38.70000076293945, "explorer_subspace_norm": 79.3697738647461, "final_distance": 1.3376797437667847, "max_param_abs": 1.6915369033813477, "policy_loss": -0.0029535914400184993, "value_loss": 3.245455265045166}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:44.690311+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.354592800140381, "chaser_new_fields": 33.939998626708984, "chaser_partner_vision": 0.28529998660087585, "chaser_return": -2.8299989700317383, "chaser_stuck_run_steps": 2.6399998664855957, "collisions_per_episode": 0.35999998450279236, "degenerate_episodes": 89, "degenerate_fraction": 0.89, "episodes": 100, "explorer_new_fields": 57.0099983215332, "explorer_partner_vision": 0.28529998660087585, "explorer_return": 35.18499755859375, "explorer_stuck_run_steps": 0.5600000023841858, "final_distance": 5.617663383483887, "same_state_run_steps": 0.04999999701976776}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.social.seed_0003.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0003"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0004/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0004/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0004/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0004.attempt_01", "seed": 4, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.004395685515618197, "chaser_grad_norm": 0.4359506666660309, "chaser_new_fields": 51.32500076293945, "chaser_partner_vision": 0.9757500290870667, "chaser_return": 13.307501792907715, "chaser_subspace_norm": 150.7742156982422, "collisions_per_episode": 12.475000381469727, "entropy": 0.1167144626379013, "explorer_grad_norm": 1.5392605364322662, "explorer_new_fields": 62.525001525878906, "explorer_partner_vision": 0.9757500290870667, "explorer_return": 32.712501525878906, "explorer_subspace_norm": 79.2755126953125, "final_distance": 1.8760557174682617, "max_param_abs": 1.57667076587677, "policy_loss": -0.003739094465231574, "value_loss": 3.158414661884308}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:44.810284+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 1.6737258434295654, "chaser_new_fields": 26.01999855041504, "chaser_partner_vision": 0.9523999691009521, "chaser_return": 39.76400375366211, "chaser_stuck_run_steps": 0.5199999809265137, "collisions_per_episode": 41.20000076293945, "degenerate_episodes": 98, "degenerate_fraction": 0.98, "episodes": 100, "explorer_new_fields": 22.760000228881836, "explorer_partner_vision": 0.9523999691009521, "explorer_return": -43.46500015258789, "explorer_stuck_run_steps": 3.609999895095825, "final_distance": 1.6019630432128906, "same_state_run_steps": 0.07999999821186066}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.social.seed_0004.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0004"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0004/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0004/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0004/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0004.attempt_01", "seed": 4, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.004395685515618197, "chaser_grad_norm": 0.4359506666660309, "chaser_new_fields": 51.32500076293945, "chaser_partner_vision": 0.9757500290870667, "chaser_return": 13.307501792907715, "chaser_subspace_norm": 150.7742156982422, "collisions_per_episode": 12.475000381469727, "entropy": 0.1167144626379013, "explorer_grad_norm": 1.5392605364322662, "explorer_new_fields": 62.525001525878906, "explorer_partner_vision": 0.9757500290870667, "explorer_return": 32.712501525878906, "explorer_subspace_norm": 79.2755126953125, "final_distance": 1.8760557174682617, "max_param_abs": 1.57667076587677, "policy_loss": -0.003739094465231574, "value_loss": 3.158414661884308}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:44.923629+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.481812477111816, "chaser_new_fields": 33.869998931884766, "chaser_partner_vision": 0.2645999789237976, "chaser_return": -2.61099910736084, "chaser_stuck_run_steps": 2.629999876022339, "collisions_per_episode": 0.5699999928474426, "degenerate_episodes": 91, "degenerate_fraction": 0.91, "episodes": 100, "explorer_new_fields": 59.21999740600586, "explorer_partner_vision": 0.2645999789237976, "explorer_return": 38.2599983215332, "explorer_stuck_run_steps": 0.4399999976158142, "final_distance": 5.834385871887207, "same_state_run_steps": 0.04999999701976776}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.social.seed_0004.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0004"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0005/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0005/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0005/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0005.attempt_01", "seed": 5, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.002886005277595416, "chaser_grad_norm": 0.5347528606653214, "chaser_new_fields": 60.67499923706055, "chaser_partner_vision": 0.9855000376701355, "chaser_return": 16.57750129699707, "chaser_subspace_norm": 246.68600463867188, "collisions_per_episode": 13.77500057220459, "entropy": 0.06711985357105732, "explorer_grad_norm": 2.149692416191101, "explorer_new_fields": 71.4749984741211, "explorer_partner_vision": 0.9855000376701355, "explorer_return": 43.837501525878906, "explorer_subspace_norm": 87.82106018066406, "final_distance": 1.5338835716247559, "max_param_abs": 1.5635184049606323, "policy_loss": -0.004267054378942703, "value_loss": 1.5971579551696777}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:45.043449+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 1.8481124639511108, "chaser_new_fields": 25.719999313354492, "chaser_partner_vision": 0.9373998641967773, "chaser_return": 28.073993682861328, "chaser_stuck_run_steps": 0.25999999046325684, "collisions_per_episode": 30.399999618530273, "degenerate_episodes": 99, "degenerate_fraction": 0.99, "episodes": 100, "explorer_new_fields": 23.15999984741211, "explorer_partner_vision": 0.9373998641967773, "explorer_return": -36.369998931884766, "explorer_stuck_run_steps": 3.5399999618530273, "final_distance": 1.725230097770691, "same_state_run_steps": 0.05999999865889549}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.social.seed_0005.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0005"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0005/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0005/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0005/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0005.attempt_01", "seed": 5, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.002886005277595416, "chaser_grad_norm": 0.5347528606653214, "chaser_new_fields": 60.67499923706055, "chaser_partner_vision": 0.9855000376701355, "chaser_return": 16.57750129699707, "chaser_subspace_norm": 246.68600463867188, "collisions_per_episode": 13.77500057220459, "entropy": 0.06711985357105732, "explorer_grad_norm": 2.149692416191101, "explorer_new_fields": 71.4749984741211, "explorer_partner_vision": 0.9855000376701355, "explorer_return": 43.837501525878906, "explorer_subspace_norm": 87.82106018066406, "final_distance": 1.5338835716247559, "max_param_abs": 1.5635184049606323, "policy_loss": -0.004267054378942703, "value_loss": 1.5971579551696777}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:45.156580+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.357820510864258, "chaser_new_fields": 33.86000061035156, "chaser_partner_vision": 0.2691999673843384, "chaser_return": -2.8059990406036377, "chaser_stuck_run_steps": 2.629999876022339, "collisions_per_episode": 0.3999999761581421, "degenerate_episodes": 90, "degenerate_fraction": 0.9, "episodes": 100, "explorer_new_fields": 56.0099983215332, "explorer_partner_vision": 0.2691999673843384, "explorer_return": 33.619998931884766, "explorer_stuck_run_steps": 0.25999999046325684, "final_distance": 5.411197185516357, "same_state_run_steps": 0.019999999552965164}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.social.seed_0005.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0005"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0006/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0006/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0006/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0006.attempt_01", "seed": 6, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.05561700484361154, "chaser_grad_norm": 1.0297226533293724, "chaser_new_fields": 51.22500228881836, "chaser_partner_vision": 0.9850000739097595, "chaser_return": 22.880001068115234, "chaser_subspace_norm": 182.61648559570312, "collisions_per_episode": 21.0, "entropy": 0.13069424219429493, "explorer_grad_norm": 1.6008991301059723, "explorer_new_fields": 60.400001525878906, "explorer_partner_vision": 0.9850000739097595, "explorer_return": 26.950000762939453, "explorer_subspace_norm": 73.82489776611328, "final_distance": 1.4356602430343628, "max_param_abs": 1.6293209791183472, "policy_loss": -0.015071312334384857, "value_loss": 4.025597631931305}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:45.276536+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 1.640023112297058, "chaser_new_fields": 27.06999969482422, "chaser_partner_vision": 0.9590999484062195, "chaser_return": 38.13800048828125, "chaser_stuck_run_steps": 0.3100000023841858, "collisions_per_episode": 39.459999084472656, "degenerate_episodes": 98, "degenerate_fraction": 0.98, "episodes": 100, "explorer_new_fields": 22.85999870300293, "explorer_partner_vision": 0.9590999484062195, "explorer_return": -42.30999755859375, "explorer_stuck_run_steps": 3.4099998474121094, "final_distance": 1.4202427864074707, "same_state_run_steps": 0.05999999865889549}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.social.seed_0006.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0006"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0006/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0006/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0006/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0006.attempt_01", "seed": 6, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.05561700484361154, "chaser_grad_norm": 1.0297226533293724, "chaser_new_fields": 51.22500228881836, "chaser_partner_vision": 0.9850000739097595, "chaser_return": 22.880001068115234, "chaser_subspace_norm": 182.61648559570312, "collisions_per_episode": 21.0, "entropy": 0.13069424219429493, "explorer_grad_norm": 1.6008991301059723, "explorer_new_fields": 60.400001525878906, "explorer_partner_vision": 0.9850000739097595, "explorer_return": 26.950000762939453, "explorer_subspace_norm": 73.82489776611328, "final_distance": 1.4356602430343628, "max_param_abs": 1.6293209791183472, "policy_loss": -0.015071312334384857, "value_loss": 4.025597631931305}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:45.389938+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.489194869995117, "chaser_new_fields": 34.0, "chaser_partner_vision": 0.2618999779224396, "chaser_return": -2.663999080657959, "chaser_stuck_run_steps": 2.619999885559082, "collisions_per_episode": 0.5, "degenerate_episodes": 88, "degenerate_fraction": 0.88, "episodes": 100, "explorer_new_fields": 64.8699951171875, "explorer_partner_vision": 0.2618999779224396, "explorer_return": 46.69499969482422, "explorer_stuck_run_steps": 0.1899999976158142, "final_distance": 5.930675983428955, "same_state_run_steps": 0.009999999776482582}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.social.seed_0006.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0006"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0007/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0007/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0007/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0007.attempt_01", "seed": 7, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.001655151105467212, "chaser_grad_norm": 0.509749636054039, "chaser_new_fields": 48.775001525878906, "chaser_partner_vision": 0.968500018119812, "chaser_return": 16.532501220703125, "chaser_subspace_norm": 136.35047912597656, "collisions_per_episode": 15.824999809265137, "entropy": 0.1547732874751091, "explorer_grad_norm": 1.284228354692459, "explorer_new_fields": 59.85000228881836, "explorer_partner_vision": 0.968500018119812, "explorer_return": 28.075000762939453, "explorer_subspace_norm": 77.23289489746094, "final_distance": 1.4770816564559937, "max_param_abs": 1.5285195112228394, "policy_loss": -0.0031741416505610687, "value_loss": 3.588924825191498}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:45.509893+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 1.632234811782837, "chaser_new_fields": 24.229999542236328, "chaser_partner_vision": 0.9589999914169312, "chaser_return": 37.8489990234375, "chaser_stuck_run_steps": 0.07000000029802322, "collisions_per_episode": 39.61000061035156, "degenerate_episodes": 99, "degenerate_fraction": 0.99, "episodes": 100, "explorer_new_fields": 21.170000076293945, "explorer_partner_vision": 0.9589999914169312, "explorer_return": -45.099998474121094, "explorer_stuck_run_steps": 3.5199999809265137, "final_distance": 1.4150844812393188, "same_state_run_steps": 0.019999999552965164}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.social.seed_0007.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0007"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0007/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0007/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0007/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0007.attempt_01", "seed": 7, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.001655151105467212, "chaser_grad_norm": 0.509749636054039, "chaser_new_fields": 48.775001525878906, "chaser_partner_vision": 0.968500018119812, "chaser_return": 16.532501220703125, "chaser_subspace_norm": 136.35047912597656, "collisions_per_episode": 15.824999809265137, "entropy": 0.1547732874751091, "explorer_grad_norm": 1.284228354692459, "explorer_new_fields": 59.85000228881836, "explorer_partner_vision": 0.968500018119812, "explorer_return": 28.075000762939453, "explorer_subspace_norm": 77.23289489746094, "final_distance": 1.4770816564559937, "max_param_abs": 1.5285195112228394, "policy_loss": -0.0031741416505610687, "value_loss": 3.588924825191498}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:45.623197+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.699781894683838, "chaser_new_fields": 33.939998626708984, "chaser_partner_vision": 0.24239997565746307, "chaser_return": -2.8149986267089844, "chaser_stuck_run_steps": 2.619999885559082, "collisions_per_episode": 0.3700000047683716, "degenerate_episodes": 89, "degenerate_fraction": 0.89, "episodes": 100, "explorer_new_fields": 63.87999725341797, "explorer_partner_vision": 0.24239997565746307, "explorer_return": 45.40999984741211, "explorer_stuck_run_steps": 0.6599999666213989, "final_distance": 5.8027777671813965, "same_state_run_steps": 0.04999999701976776}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.social.seed_0007.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0007"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0008/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0008/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0008/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0008.attempt_01", "seed": 8, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.0029020164416238003, "chaser_grad_norm": 0.6062818244099617, "chaser_new_fields": 59.45000076293945, "chaser_partner_vision": 0.9830000400543213, "chaser_return": 16.53999900817871, "chaser_subspace_norm": 186.8477325439453, "collisions_per_episode": 13.949999809265137, "entropy": 0.0607997989282012, "explorer_grad_norm": 1.2333895713090897, "explorer_new_fields": 76.42500305175781, "explorer_partner_vision": 0.9830000400543213, "explorer_return": 49.82500076293945, "explorer_subspace_norm": 70.7031021118164, "final_distance": 1.571839451789856, "max_param_abs": 1.6751644611358643, "policy_loss": -0.0022539286062510655, "value_loss": 1.0913494527339935}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:45.743025+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 1.928891897201538, "chaser_new_fields": 25.079999923706055, "chaser_partner_vision": 0.9176999926567078, "chaser_return": 29.802997589111328, "chaser_stuck_run_steps": 0.0, "collisions_per_episode": 32.130001068115234, "degenerate_episodes": 94, "degenerate_fraction": 0.94, "episodes": 100, "explorer_new_fields": 24.15999984741211, "explorer_partner_vision": 0.9176999926567078, "explorer_return": -35.67499923706055, "explorer_stuck_run_steps": 3.429999828338623, "final_distance": 1.6439684629440308, "same_state_run_steps": 0.0}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.social.seed_0008.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0008"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0008/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0008/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0008/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0008.attempt_01", "seed": 8, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.0029020164416238003, "chaser_grad_norm": 0.6062818244099617, "chaser_new_fields": 59.45000076293945, "chaser_partner_vision": 0.9830000400543213, "chaser_return": 16.53999900817871, "chaser_subspace_norm": 186.8477325439453, "collisions_per_episode": 13.949999809265137, "entropy": 0.0607997989282012, "explorer_grad_norm": 1.2333895713090897, "explorer_new_fields": 76.42500305175781, "explorer_partner_vision": 0.9830000400543213, "explorer_return": 49.82500076293945, "explorer_subspace_norm": 70.7031021118164, "final_distance": 1.571839451789856, "max_param_abs": 1.6751644611358643, "policy_loss": -0.0022539286062510655, "value_loss": 1.0913494527339935}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:45.856335+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.189486980438232, "chaser_new_fields": 33.97999954223633, "chaser_partner_vision": 0.2847999930381775, "chaser_return": -2.90799880027771, "chaser_stuck_run_steps": 2.619999885559082, "collisions_per_episode": 0.2800000011920929, "degenerate_episodes": 89, "degenerate_fraction": 0.89, "episodes": 100, "explorer_new_fields": 44.84000015258789, "explorer_partner_vision": 0.2847999930381775, "explorer_return": 17.014999389648438, "explorer_stuck_run_steps": 0.6299999952316284, "final_distance": 5.741815090179443, "same_state_run_steps": 0.03999999910593033}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.social.seed_0008.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0008"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0009/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0009/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0009/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0009.attempt_01", "seed": 9, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.00037949177203699946, "chaser_grad_norm": 3.413390338420868, "chaser_new_fields": 57.67499923706055, "chaser_partner_vision": 0.9892500638961792, "chaser_return": 21.462499618530273, "chaser_subspace_norm": 188.78103637695312, "collisions_per_episode": 18.524999618530273, "entropy": 0.08403550647199154, "explorer_grad_norm": 5.088832139968872, "explorer_new_fields": 69.3499984741211, "explorer_partner_vision": 0.9892500638961792, "explorer_return": 40.1875, "explorer_subspace_norm": 97.37679290771484, "final_distance": 1.3710155487060547, "max_param_abs": 1.6251237392425537, "policy_loss": -0.0006575217575459646, "value_loss": 2.170237898826599}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:45.976230+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "chaser", "max_steps": 100, "metrics": {"average_distance": 1.8344014883041382, "chaser_new_fields": 26.40999984741211, "chaser_partner_vision": 0.9383999705314636, "chaser_return": 31.125, "chaser_stuck_run_steps": 0.5199999809265137, "collisions_per_episode": 33.14999771118164, "degenerate_episodes": 95, "degenerate_fraction": 0.95, "episodes": 100, "explorer_new_fields": 23.939998626708984, "explorer_partner_vision": 0.9383999705314636, "explorer_return": -37.040000915527344, "explorer_stuck_run_steps": 3.5199999809265137, "final_distance": 1.6236257553100586, "same_state_run_steps": 0.23999999463558197}, "opponent_mode": "random_explorer", "run_id": "mouse-run-run-2-ssm.social.seed_0009.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0009"} | |
| {"attempt_id": "attempt_01", "batch_size": 128, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0009/attempt_01/checkpoints/latest.safetensors", "checkpoint_config": {"architecture": "ssm", "attempt_id": "attempt_01", "batch_size": 40, "checkpoint": "runs/mouse-run-run-2-ssm/social/seed_0009/attempt_01/checkpoints/latest.safetensors", "checkpoint_every": 1000, "checkpoint_every_seconds": 1800.0, "clip_epsilon": 0.2, "cost_per_hour": null, "cuda_tf32": true, "device": "cuda", "entropy_coef": 0.01, "env": {"grid_size": 10, "max_steps": 100, "partner_visibility": "partial", "task": "social", "vision_radius": 3}, "experiment_id": "mouse-run-run-2-ssm", "finite_guard": true, "fused_agent_rollout": true, "gae_lambda": 0.95, "gamma": 0.99, "grad_clip": 1.0, "hidden_size": 256, "learning_rate": 0.0003, "log_every": 500, "metrics_path": null, "ppo_epochs": 4, "recurrent_l2_coef": 0.3, "resume_from": null, "run_dir": "runs/mouse-run-run-2-ssm/social/seed_0009/attempt_01", "run_id": "mouse-run-run-2-ssm.social.seed_0009.attempt_01", "seed": 9, "status_every_seconds": 1800.0, "status_path": null, "subspace_metric_period": 10, "tensorboard_dir": null, "triton_env_step": true, "updates": 20000, "value_clip": 10.0, "value_coef": 0.5}, "checkpoint_health": {"failures": [], "format_ok": true, "max_abs": 80000.0, "nonfinite_tensor_count": 0, "ok": true, "tensor_count": 130}, "checkpoint_metrics": {"approx_kl": 0.00037949177203699946, "chaser_grad_norm": 3.413390338420868, "chaser_new_fields": 57.67499923706055, "chaser_partner_vision": 0.9892500638961792, "chaser_return": 21.462499618530273, "chaser_subspace_norm": 188.78103637695312, "collisions_per_episode": 18.524999618530273, "entropy": 0.08403550647199154, "explorer_grad_norm": 5.088832139968872, "explorer_new_fields": 69.3499984741211, "explorer_partner_vision": 0.9892500638961792, "explorer_return": 40.1875, "explorer_subspace_norm": 97.37679290771484, "final_distance": 1.3710155487060547, "max_param_abs": 1.6251237392425537, "policy_loss": -0.0006575217575459646, "value_loss": 2.170237898826599}, "command": "/root/mouse-run-run-perf/.venv/bin/evaluate-paper-marl runs/mouse-run-run-2-ssm --output runs/mouse-run-run-2-ssm/paper_random_eval.jsonl --device cuda", "created_at": "2026-07-03T11:34:46.089552+00:00", "degenerate_threshold_fraction": 0.01, "deterministic": false, "device": "cuda", "episodes": 100, "error": null, "evaluation_protocol": "paper_random_opponent_v1", "experiment_id": "mouse-run-run-2-ssm", "focus_agent": "explorer", "max_steps": 100, "metrics": {"average_distance": 5.345434665679932, "chaser_new_fields": 33.98999786376953, "chaser_partner_vision": 0.28029999136924744, "chaser_return": -2.607998847961426, "chaser_stuck_run_steps": 2.609999895095825, "collisions_per_episode": 0.5600000023841858, "degenerate_episodes": 92, "degenerate_fraction": 0.92, "episodes": 100, "explorer_new_fields": 53.25, "explorer_partner_vision": 0.28029999136924744, "explorer_return": 29.309999465942383, "explorer_stuck_run_steps": 1.809999942779541, "final_distance": 5.206815242767334, "same_state_run_steps": 0.19999998807907104}, "opponent_mode": "random_chaser", "run_id": "mouse-run-run-2-ssm.social.seed_0009.attempt_01", "schema_version": 3, "seed": 0, "status": "ok", "task": "social", "unit_id": "social/seed_0009"} | |