Retain archived profile-trained episode evidence beside both mean/profile comparison conditions

evaluations/deadly-corridor/deadly_corridor-qwenoft-hist8-scratch-s42-v2/eval_profile.json

@@ -0,0 +1,44 @@
+{
+  "job_id": "eval_v2__deadly_corridor-qwenoft-hist8-scratch-s42-v2",
+  "eval_protocol_version": "hist8-history01-sf-render-v4",
+  "training_step": 5000,
+  "checkpoint_sha256": "29e42431b2fb4e83e74f21c0e7e108d43ebbeb91ab7c34da06c33cb0652fb56c",
+  "profile_sha256": "bbfb93cef9f63750a9a159ec10a93a9fc6c25e4cb0cac3b39532663b4dc11eba",
+  "mean_return": 2091.234108123779,
+  "std_return": 579.6123067681043,
+  "mean_length": 174.1,
+  "episodes": 100,
+  "parallel_envs": 32,
+  "max_episode_raw_frames": 3600,
+  "issued_actions": 4390,
+  "top_action": "MOVE_FORWARD+MOVE_RIGHT+ATTACK",
+  "largest_action_fraction": 0.3323462414578588,
+  "unique_action_values": 16,
+  "episode_final_done": 100,
+  "episodes_at_frame_cap": 0,
+  "checks": {
+    "100_exact_seeds": true,
+    "checkpoint_sha": true,
+    "bundle_step": true,
+    "training_step_matches_plan": true,
+    "profile_sha": true,
+    "binary_history": true,
+    "raw_score_means_match": true,
+    "action_counts_match": true,
+    "action_total_match": true,
+    "summary_returns_match": true,
+    "summary_lengths_match": true,
+    "100_final_steps": true,
+    "final_step_lengths_match": true,
+    "renderer_matches_training": true,
+    "renderer_protocol_v4": true,
+    "parallel_capacity_horizon": true,
+    "fps_history": true,
+    "environment_latency_matches": true,
+    "episode_latency_protocol": true,
+    "training_recipe": true,
+    "training_dimensions": true,
+    "initialization_revision": true
+  },
+  "environment_condition": "profile"
+}

evaluations/deadly-corridor/deadly_corridor-qwenoft-hist8-scratch-s42-v2/evaluations/profile/action_audit.json

@@ -0,0 +1,154 @@
+{
+  "eval_protocol_version": "hist8-history01-sf-render-v4",
+  "history_encoding": "binary_0_1",
+  "episodes": 100,
+  "episode_seeds": [
+    1000000,
+    1000001,
+    1000002,
+    1000003,
+    1000004,
+    1000005,
+    1000006,
+    1000007,
+    1000008,
+    1000009,
+    1000010,
+    1000011,
+    1000012,
+    1000013,
+    1000014,
+    1000015,
+    1000016,
+    1000017,
+    1000018,
+    1000019,
+    1000020,
+    1000021,
+    1000022,
+    1000023,
+    1000024,
+    1000025,
+    1000026,
+    1000027,
+    1000028,
+    1000029,
+    1000030,
+    1000031,
+    1000032,
+    1000033,
+    1000034,
+    1000035,
+    1000036,
+    1000037,
+    1000038,
+    1000039,
+    1000040,
+    1000041,
+    1000042,
+    1000043,
+    1000044,
+    1000045,
+    1000046,
+    1000047,
+    1000048,
+    1000049,
+    1000050,
+    1000051,
+    1000052,
+    1000053,
+    1000054,
+    1000055,
+    1000056,
+    1000057,
+    1000058,
+    1000059,
+    1000060,
+    1000061,
+    1000062,
+    1000063,
+    1000064,
+    1000065,
+    1000066,
+    1000067,
+    1000068,
+    1000069,
+    1000070,
+    1000071,
+    1000072,
+    1000073,
+    1000074,
+    1000075,
+    1000076,
+    1000077,
+    1000078,
+    1000079,
+    1000080,
+    1000081,
+    1000082,
+    1000083,
+    1000084,
+    1000085,
+    1000086,
+    1000087,
+    1000088,
+    1000089,
+    1000090,
+    1000091,
+    1000092,
+    1000093,
+    1000094,
+    1000095,
+    1000096,
+    1000097,
+    1000098,
+    1000099
+  ],
+  "issued_actions": 4390,
+  "action_counts": {
+    "[1,0,0,1,1,0,1]": 730,
+    "[1,0,0,1,0,0,1]": 1459,
+    "[0,0,0,1,0,1,1]": 53,
+    "[1,0,0,1,0,1,1]": 361,
+    "[1,0,0,0,0,0,1]": 604,
+    "[1,0,0,0,0,0,0]": 142,
+    "[1,0,1,0,0,0,0]": 348,
+    "[1,0,1,0,0,0,1]": 225,
+    "[1,0,1,0,1,0,0]": 102,
+    "[1,0,0,0,1,0,1]": 90,
+    "[1,0,0,1,0,0,0]": 44,
+    "[1,0,0,1,1,0,0]": 105,
+    "[0,0,0,1,0,0,1]": 5,
+    "[1,0,0,0,0,1,1]": 34,
+    "[1,0,1,0,1,0,1]": 72,
+    "[1,0,0,0,1,0,0]": 16
+  },
+  "largest_action_fraction": 0.3323462414578588,
+  "raw_score_min": [
+    -13.655267715454102,
+    -35.56375503540039,
+    -17.454011917114258,
+    -17.415475845336914,
+    -17.392902374267578,
+    -16.934200286865234,
+    -13.61490535736084
+  ],
+  "raw_score_max": [
+    16.83314323425293,
+    -12.029019355773926,
+    15.080259323120117,
+    19.629093170166016,
+    16.658203125,
+    16.210956573486328,
+    16.159494400024414
+  ],
+  "raw_score_mean": [
+    11.341447180779417,
+    -25.805974498262167,
+    -8.314430420248343,
+    3.713643527584223,
+    -5.694479129193584,
+    -9.024148428242373,
+    6.775425593843068
+  ]
+}

evaluations/deadly-corridor/deadly_corridor-qwenoft-hist8-scratch-s42-v2/evaluations/profile/eval_config.yaml

@@ -0,0 +1,166 @@
+experiment:
+  name: deadly_corridor-qwenoft-hist8-scratch-s42-v2-step5000-history01-profile-par32
+  seed: 1000000
+backend:
+  type: sample_factory
+  algo: APPO
+  device: cpu
+  train_dir: results/sample_factory
+  restart_behavior: resume
+  run_mode: eval
+executor:
+  mode: simulated
+  simulated_worker_capacity: 1
+  simulated_inference_pool: true
+  inference_devices:
+  - cuda:0
+  - cuda:1
+  inference_batch_size: 16
+env:
+  name: deadly_corridor
+  env_id: doom_deadly_corridor
+  env_fps: 35
+  obs_fps: 8.75
+  noop_action:
+  - 0
+  - 0
+  - 0
+  - 0
+  frame_stack: 1
+  res_w: 128
+  res_h: 72
+  wide_aspect_ratio: false
+  simulator: cpu
+  obs_resize:
+  - 224
+  - 224
+  action_map:
+    noop:
+    - 0
+    - 0
+    - 0
+    - 0
+    move_forward:
+    - 0
+    - 1
+    - 0
+    - 0
+    move_backward:
+    - 0
+    - 2
+    - 0
+    - 0
+    move_left:
+    - 0
+    - 0
+    - 1
+    - 0
+    move_right:
+    - 0
+    - 0
+    - 2
+    - 0
+    turn_left:
+    - 1
+    - 0
+    - 0
+    - 0
+    turn_right:
+    - 2
+    - 0
+    - 0
+    - 0
+    attack:
+    - 0
+    - 0
+    - 0
+    - 1
+  action_history_decisions: 8
+  screen_resolution: RES_160X120
+  render_hud: true
+  render_crosshair: false
+  render_weapon: true
+  render_decals: false
+  render_particles: false
+latency:
+  method: temporal
+  profile_path: profiles/profile.json
+  profile_worker_slot: 0
+  seed: 271828
+  add_latency_info: false
+scheduler:
+  hold_policy: hold
+  ordering_policy: latest_ready
+policy:
+  type: starvla
+  actions:
+  - MOVE_FORWARD
+  - MOVE_BACKWARD
+  - MOVE_LEFT
+  - MOVE_RIGHT
+  - TURN_LEFT
+  - TURN_RIGHT
+  - ATTACK
+  checkpoint_path: checkpoints/model.pt
+  model_config_path: config.yaml
+  device: cuda:0
+  unnorm_key: new_embodiment
+  prompt_mode: latency_neutral
+  action_layout: multibinary_7
+  backbone_path: BACKBONE_SNAPSHOT
+  worker_python_executable: python
+  state_source: transport
+training:
+  train_for_env_steps: 25000000
+  num_workers: 32
+  num_envs_per_worker: 4
+  worker_num_splits: 2
+  num_policies: 1
+  batch_size: 1024
+  rollout: 128
+  recurrence: 128
+  num_epochs: 2
+  num_batches_per_epoch: 2
+  learning_rate: 0.0001
+  gamma: 0.99
+  gae_lambda: 0.95
+  ppo_clip_ratio: 0.1
+  ppo_clip_value: 0.2
+  exploration_loss: symmetric_kl
+  exploration_loss_coeff: 0.001
+  value_loss_coeff: 0.5
+  max_grad_norm: 4.0
+  async_rl: true
+  use_rnn: true
+  rnn_type: gru
+  rnn_size: 512
+  normalize_input: true
+  normalize_returns: true
+  stats_avg: 100
+  experiment_summaries_interval: 1
+  save_every_sec: 600
+  keep_checkpoints: 5
+evaluation:
+  eval_interval_steps: 1000000
+  eval_episodes: 100
+  eval_parallel_envs: 32
+  eval_max_steps: 3600
+  eval_deterministic: true
+  eval_raw_reward: true
+  eval_suites:
+    fixed: []
+    normal: []
+    uniform: []
+  eval_latency_values: null
+logging:
+  output_dir: replay_outputs/profile
+  video:
+    enabled: false
+    num_bins: 10
+  save_step_records: true
+  save_action_records: true
+  save_latency_records: true
+  wandb_project: null
+  wandb_group: null
+  wandb_job_type: null
+  wandb_tags: ''

evaluations/deadly-corridor/deadly_corridor-qwenoft-hist8-scratch-s42-v2/evaluations/profile/provenance.json

@@ -0,0 +1,35 @@
+{
+  "collection": {
+    "teacher_repo": "latency-sensitive-bench/benchmark-models",
+    "teacher_revision": "af125bd372c22c5b00e326d9af1af64ea3751b56",
+    "teacher_path": "latency-aware/deadly-corridor/small-policy/sample-factory-qwenoft-rtx3090-temporal-hist8-capacity1-native35-s42-v1",
+    "teacher_checkpoint": "best_000044776_22925312_reward_2084.409.pth",
+    "teacher_sha256": "62eee5b587bde20cc912fde9c068d92498a6433b62dec28f8e994004fb8e7e18",
+    "profile_path": "source_archive/hist8-vla-20260923/results/profiling/profiles/openvla/1x-rtx3090/deadly_corridor/instance_a5037b165aa0cedc/profile.json",
+    "profile_sha256": "bbfb93cef9f63750a9a159ec10a93a9fc6c25e4cb0cac3b39532663b4dc11eba",
+    "episodes": 8500,
+    "filter_preset": "none",
+    "history": "decision_v1"
+  },
+  "training_run_id": "deadly_corridor-qwenoft-hist8-scratch-s42-v2",
+  "training_step": 5000,
+  "global_batch": 128,
+  "base_model": "Qwen/Qwen3-VL-4B-Instruct",
+  "wandb_run_id": "1ade216c",
+  "checkpoint_sha256": "29e42431b2fb4e83e74f21c0e7e108d43ebbeb91ab7c34da06c33cb0652fb56c",
+  "eval_protocol_version": "hist8-history01-sf-render-v4",
+  "history_encoding": "binary_0_1",
+  "episodes": 100,
+  "env_condition": "profile",
+  "parallel_envs": 32,
+  "max_episode_raw_frames": 3600,
+  "inference_devices": [
+    "cuda:0",
+    "cuda:1"
+  ],
+  "episode_seeds": [
+    1000000,
+    1000099
+  ],
+  "profile_sha256": "bbfb93cef9f63750a9a159ec10a93a9fc6c25e4cb0cac3b39532663b4dc11eba"
+}

evaluations/deadly-corridor/deadly_corridor-qwenoft-hist8-scratch-s42-v2/evaluations/profile/resolved_config.yaml

@@ -0,0 +1,169 @@
+experiment:
+  name: deadly_corridor-qwenoft-hist8-scratch-s42-v2-step5000-history01-profile-par32
+  seed: 1000000
+backend:
+  type: sample_factory
+  algo: APPO
+  device: cpu
+  train_dir: results/sample_factory
+  restart_behavior: resume
+  run_mode: eval
+executor:
+  mode: simulated
+  simulated_worker_capacity: 1
+  simulated_inference_pool: true
+  inference_devices:
+  - cuda:0
+  - cuda:1
+  inference_batch_size: 16
+env:
+  name: deadly_corridor
+  env_id: doom_deadly_corridor
+  env_fps: 35
+  obs_fps: 8.75
+  noop_action:
+  - 0
+  - 0
+  - 0
+  - 0
+  frame_stack: 1
+  res_w: 128
+  res_h: 72
+  wide_aspect_ratio: false
+  simulator: cpu
+  obs_resize:
+  - 224
+  - 224
+  action_map:
+    noop:
+    - 0
+    - 0
+    - 0
+    - 0
+    move_forward:
+    - 0
+    - 1
+    - 0
+    - 0
+    move_backward:
+    - 0
+    - 2
+    - 0
+    - 0
+    move_left:
+    - 0
+    - 0
+    - 1
+    - 0
+    move_right:
+    - 0
+    - 0
+    - 2
+    - 0
+    turn_left:
+    - 1
+    - 0
+    - 0
+    - 0
+    turn_right:
+    - 2
+    - 0
+    - 0
+    - 0
+    attack:
+    - 0
+    - 0
+    - 0
+    - 1
+  action_history_decisions: 8
+  screen_resolution: RES_160X120
+  render_hud: true
+  render_crosshair: false
+  render_weapon: true
+  render_decals: false
+  render_particles: false
+latency:
+  method: temporal
+  profile_path: profiles/profile.json
+  profile_worker_slot: 0
+  seed: 271828
+  add_latency_info: false
+scheduler:
+  hold_policy: hold
+  ordering_policy: latest_ready
+policy:
+  type: starvla
+  actions:
+  - MOVE_FORWARD
+  - MOVE_BACKWARD
+  - MOVE_LEFT
+  - MOVE_RIGHT
+  - TURN_LEFT
+  - TURN_RIGHT
+  - ATTACK
+  checkpoint_path: checkpoints/model.pt
+  model_config_path: config.yaml
+  device: cuda:0
+  unnorm_key: new_embodiment
+  prompt_mode: latency_neutral
+  action_layout: multibinary_7
+  backbone_path: BACKBONE_SNAPSHOT
+  worker_python_executable: python
+  state_source: transport
+  action_prefix:
+    mode: none
+training:
+  train_for_env_steps: 25000000
+  num_workers: 32
+  num_envs_per_worker: 4
+  worker_num_splits: 2
+  num_policies: 1
+  batch_size: 1024
+  rollout: 128
+  recurrence: 128
+  num_epochs: 2
+  num_batches_per_epoch: 2
+  learning_rate: 0.0001
+  gamma: 0.99
+  gae_lambda: 0.95
+  ppo_clip_ratio: 0.1
+  ppo_clip_value: 0.2
+  exploration_loss: symmetric_kl
+  exploration_loss_coeff: 0.001
+  value_loss_coeff: 0.5
+  max_grad_norm: 4.0
+  async_rl: true
+  use_rnn: true
+  rnn_type: gru
+  rnn_size: 512
+  normalize_input: true
+  normalize_returns: true
+  stats_avg: 100
+  experiment_summaries_interval: 1
+  save_every_sec: 600
+  keep_checkpoints: 5
+evaluation:
+  eval_interval_steps: 1000000
+  eval_episodes: 100
+  eval_parallel_envs: 32
+  eval_max_steps: 3600
+  eval_deterministic: true
+  eval_raw_reward: true
+  eval_suites:
+    fixed: []
+    normal: []
+    uniform: []
+  eval_latency_values: null
+logging:
+  output_dir: replay_outputs/profile
+  video:
+    enabled: false
+    num_bins: 10
+  save_step_records: true
+  save_action_records: true
+  save_latency_records: true
+  wandb_project: null
+  wandb_group: null
+  wandb_job_type: null
+  wandb_tags: ''
+  simulated_pipeline_profile: false

evaluations/deadly-corridor/deadly_corridor-qwenoft-hist8-scratch-s42-v2/evaluations/profile/summary.json

@@ -0,0 +1,217 @@
+{
+  "checkpoint_path": "source_archive/hist8-vla-v2-20260923/outputs/tasks/decision_history_vla_hist8_v2/bundles/deadly_corridor-qwenoft-hist8-scratch-s42-v2/checkpoints/model.pt",
+  "experiment_name": "deadly_corridor-qwenoft-hist8-scratch-s42-v2-step5000-history01-profile-par32",
+  "latency": "profile_sample",
+  "latency_type": "profile_sample",
+  "lengths": [
+    185,
+    177,
+    115,
+    182,
+    19,
+    181,
+    180,
+    188,
+    175,
+    178,
+    186,
+    178,
+    181,
+    180,
+    186,
+    178,
+    190,
+    184,
+    184,
+    177,
+    189,
+    183,
+    180,
+    189,
+    187,
+    182,
+    185,
+    48,
+    186,
+    187,
+    181,
+    184,
+    186,
+    186,
+    187,
+    184,
+    180,
+    184,
+    180,
+    100,
+    184,
+    185,
+    184,
+    70,
+    184,
+    186,
+    185,
+    11,
+    133,
+    142,
+    180,
+    146,
+    185,
+    184,
+    181,
+    177,
+    187,
+    196,
+    185,
+    185,
+    185,
+    191,
+    184,
+    184,
+    186,
+    185,
+    179,
+    186,
+    190,
+    193,
+    185,
+    185,
+    190,
+    182,
+    184,
+    183,
+    186,
+    184,
+    190,
+    180,
+    189,
+    184,
+    184,
+    178,
+    175,
+    184,
+    179,
+    177,
+    185,
+    203,
+    183,
+    184,
+    71,
+    191,
+    184,
+    185,
+    177,
+    175,
+    185,
+    183
+  ],
+  "mean_length": 174.1,
+  "mean_return": 2091.234108123779,
+  "returns": [
+    2280.349655151367,
+    2279.897476196289,
+    638.1345825195312,
+    2287.327117919922,
+    -66.8345947265625,
+    2281.04931640625,
+    2286.0109252929688,
+    2276.727005004883,
+    2279.5109252929688,
+    2285.495574951172,
+    2281.308639526367,
+    2276.806625366211,
+    2286.3915100097656,
+    2280.638900756836,
+    2277.4698944091797,
+    2283.752716064453,
+    2277.472442626953,
+    2276.4445190429688,
+    2276.836166381836,
+    2280.754104614258,
+    2276.6298217773438,
+    2280.7109375,
+    2283.556686401367,
+    2286.936813354492,
+    2286.5198822021484,
+    2277.2221069335938,
+    2280.7076416015625,
+    90.695556640625,
+    2280.109085083008,
+    2278.6051177978516,
+    2281.3623809814453,
+    2286.251419067383,
+    2286.775436401367,
+    2283.134002685547,
+    2281.788330078125,
+    2277.0347747802734,
+    2279.7404174804688,
+    2277.7116088867188,
+    2285.951950073242,
+    468.4805908203125,
+    2280.8585052490234,
+    2283.918472290039,
+    2283.581344604492,
+    232.80880737304688,
+    2277.3789825439453,
+    2284.1030883789062,
+    2280.2661743164062,
+    -100.53115844726562,
+    748.5314331054688,
+    692.0784301757812,
+    2281.3665924072266,
+    793.3515930175781,
+    2279.5313720703125,
+    2279.076461791992,
+    2283.1058349609375,
+    2283.738296508789,
+    2282.5528411865234,
+    2276.1557006835938,
+    2279.8590545654297,
+    2286.4039459228516,
+    2281.5637969970703,
+    2282.1121215820312,
+    2279.812789916992,
+    2280.6983337402344,
+    2277.2559967041016,
+    2281.335220336914,
+    2281.8287506103516,
+    2284.6966247558594,
+    2281.581268310547,
+    2280.741195678711,
+    2284.4117584228516,
+    2276.1195220947266,
+    2282.575881958008,
+    2278.5982360839844,
+    2279.5038146972656,
+    2287.158157348633,
+    2285.319580078125,
+    2287.1574096679688,
+    2278.002716064453,
+    2277.7862396240234,
+    2284.1024017333984,
+    2284.786895751953,
+    2283.01318359375,
+    2286.0978393554688,
+    2276.8072509765625,
+    2281.360107421875,
+    2283.258041381836,
+    2285.5669860839844,
+    2281.1952209472656,
+    2277.268341064453,
+    2278.0280151367188,
+    2284.026092529297,
+    293.99639892578125,
+    2278.020965576172,
+    2286.3648834228516,
+    2281.907760620117,
+    2285.904739379883,
+    2278.1949768066406,
+    2282.615280151367,
+    2279.032180786133
+  ],
+  "seed": 1000000,
+  "source_profile_path": "source_archive/hist8-vla-20260923/results/profiling/profiles/openvla/1x-rtx3090/deadly_corridor/instance_a5037b165aa0cedc/profile.json",
+  "std_return": 579.6123067681043,
+  "suite_name": "profile_sample",
+  "timestamp_utc": "2026-09-23T23:56:33.931684+00:00"
+}

evaluations/deadly-corridor/deadly_corridor-qwenoft-hist8-scratch-s42-v2/evaluations/profile/verification.json

@@ -0,0 +1,70 @@
+{
+  "verification_status": "verified",
+  "job_id": "eval_v2__deadly_corridor-qwenoft-hist8-scratch-s42-v2",
+  "task": "deadly_corridor",
+  "model": "qwenoft",
+  "remote_eval_dir": "source_archive/hist8-vla-v2-20260923/outputs/tasks/decision_history_vla_hist8_v2/evals/deadly_corridor-qwenoft-hist8-scratch-s42-v2-to-profile-par32-sf-render-v4",
+  "checked_at": 1790207994.4064398,
+  "training_step": 5000,
+  "checkpoint_sha256": "29e42431b2fb4e83e74f21c0e7e108d43ebbeb91ab7c34da06c33cb0652fb56c",
+  "profile_sha256": "bbfb93cef9f63750a9a159ec10a93a9fc6c25e4cb0cac3b39532663b4dc11eba",
+  "wandb_run_id": "1ade216c",
+  "eval_protocol_version": "hist8-history01-sf-render-v4",
+  "max_episode_raw_frames": 3600,
+  "parallel_envs": 32,
+  "mean_return": 2091.234108123779,
+  "std_return": 579.6123067681043,
+  "mean_length": 174.1,
+  "episodes": 100,
+  "issued_actions": 4390,
+  "top_action": "MOVE_FORWARD+MOVE_RIGHT+ATTACK",
+  "largest_action_fraction": 0.3323462414578588,
+  "unique_action_values": 16,
+  "episode_final_done": 100,
+  "episodes_at_frame_cap": 0,
+  "checks": {
+    "100_exact_seeds": true,
+    "checkpoint_sha": true,
+    "bundle_step": true,
+    "training_step_matches_plan": true,
+    "profile_sha": true,
+    "binary_history": true,
+    "raw_score_means_match": true,
+    "action_counts_match": true,
+    "action_total_match": true,
+    "summary_returns_match": true,
+    "summary_lengths_match": true,
+    "100_final_steps": true,
+    "final_step_lengths_match": true,
+    "renderer_matches_training": true,
+    "renderer_protocol_v4": true,
+    "parallel_capacity_horizon": true,
+    "fps_history": true,
+    "environment_latency_matches": true,
+    "episode_latency_protocol": true,
+    "training_recipe": true,
+    "training_dimensions": true,
+    "initialization_revision": true
+  },
+  "job": {
+    "id": "eval_v2__deadly_corridor-qwenoft-hist8-scratch-s42-v2",
+    "source": "v2",
+    "python": "source_archive/home/ubuntu/miniconda3/envs/starvla_rl_games_openvla/bin/python",
+    "task": "deadly_corridor",
+    "model": "qwenoft",
+    "backbone": "source_archive/hist8-vla-20260923/backbone/Qwen3-VL-4B-Instruct",
+    "profile_path": "source_archive/hist8-vla-20260923/results/profiling/profiles/openvla/1x-rtx3090/deadly_corridor/instance_a5037b165aa0cedc/profile.json",
+    "bundle_dir": "source_archive/hist8-vla-v2-20260923/outputs/tasks/decision_history_vla_hist8_v2/bundles/deadly_corridor-qwenoft-hist8-scratch-s42-v2",
+    "priority": 1000,
+    "depends_on": [],
+    "kind": "eval_ready",
+    "runtime_root": "source_archive/hist8-vla-eval-sf-render-v4-20260924",
+    "env_condition": "profile",
+    "parallel_envs": 32,
+    "metric_prefix": "eval_hist8_v4/profile_to_profile",
+    "eval_dir": "source_archive/hist8-vla-v2-20260923/outputs/tasks/decision_history_vla_hist8_v2/evals/deadly_corridor-qwenoft-hist8-scratch-s42-v2-to-profile-par32-sf-render-v4",
+    "gpus_required": 2,
+    "ready_path": "source_archive/hist8-vla-v2-20260923/outputs/tasks/decision_history_vla_hist8_v2/bundles/deadly_corridor-qwenoft-hist8-scratch-s42-v2/bundle.complete.json"
+  },
+  "evidence_dir": "source_archive/outputs/tasks/decision_history_vla_hist8_v2/evaluation_receipts/h100/eval_v2__deadly_corridor-qwenoft-hist8-scratch-s42-v2/hist8-history01-sf-render-v4"
+}

evaluations/flappy/flappy-qwenoft-hist8-scratch-s42-v2/eval_profile.json

@@ -0,0 +1,41 @@
+{
+  "job_id": "eval_v2__flappy-qwenoft-hist8-scratch-s42-v2",
+  "eval_protocol_version": "hist8-history01-par32-v3",
+  "training_step": 5000,
+  "checkpoint_sha256": "c6c1f36ab709d596a9123d395377d43fa57d6e9e4b8399778f3b374ca501f883",
+  "profile_sha256": "c266cba7ebac05c7e37bc83f1f16fe4b27dab97942ff43290e6c41514f6e39fc",
+  "mean_return": 373.7790043977648,
+  "std_return": 126.82699726383991,
+  "mean_length": 3030.62,
+  "episodes": 100,
+  "parallel_envs": 32,
+  "max_episode_raw_frames": 3600,
+  "issued_actions": 302228,
+  "top_action": "noop",
+  "largest_action_fraction": 0.8740189525788478,
+  "unique_action_values": 2,
+  "episode_final_done": 34,
+  "episodes_at_frame_cap": 66,
+  "checks": {
+    "100_exact_seeds": true,
+    "checkpoint_sha": true,
+    "bundle_step": true,
+    "profile_sha": true,
+    "binary_history": true,
+    "raw_score_means_match": true,
+    "action_counts_match": true,
+    "action_total_match": true,
+    "summary_returns_match": true,
+    "summary_lengths_match": true,
+    "100_final_steps": true,
+    "final_step_lengths_match": true,
+    "parallel_capacity_horizon": true,
+    "fps_history": true,
+    "environment_latency_matches": true,
+    "episode_latency_protocol": true,
+    "training_recipe": true,
+    "training_dimensions": true,
+    "initialization_revision": true
+  },
+  "environment_condition": "profile"
+}

evaluations/flappy/flappy-qwenoft-hist8-scratch-s42-v2/evaluations/profile/action_audit.json

@@ -0,0 +1,125 @@
+{
+  "eval_protocol_version": "hist8-history01-par32-v3",
+  "history_encoding": "binary_0_1",
+  "episodes": 100,
+  "episode_seeds": [
+    1000000,
+    1000001,
+    1000002,
+    1000003,
+    1000004,
+    1000005,
+    1000006,
+    1000007,
+    1000008,
+    1000009,
+    1000010,
+    1000011,
+    1000012,
+    1000013,
+    1000014,
+    1000015,
+    1000016,
+    1000017,
+    1000018,
+    1000019,
+    1000020,
+    1000021,
+    1000022,
+    1000023,
+    1000024,
+    1000025,
+    1000026,
+    1000027,
+    1000028,
+    1000029,
+    1000030,
+    1000031,
+    1000032,
+    1000033,
+    1000034,
+    1000035,
+    1000036,
+    1000037,
+    1000038,
+    1000039,
+    1000040,
+    1000041,
+    1000042,
+    1000043,
+    1000044,
+    1000045,
+    1000046,
+    1000047,
+    1000048,
+    1000049,
+    1000050,
+    1000051,
+    1000052,
+    1000053,
+    1000054,
+    1000055,
+    1000056,
+    1000057,
+    1000058,
+    1000059,
+    1000060,
+    1000061,
+    1000062,
+    1000063,
+    1000064,
+    1000065,
+    1000066,
+    1000067,
+    1000068,
+    1000069,
+    1000070,
+    1000071,
+    1000072,
+    1000073,
+    1000074,
+    1000075,
+    1000076,
+    1000077,
+    1000078,
+    1000079,
+    1000080,
+    1000081,
+    1000082,
+    1000083,
+    1000084,
+    1000085,
+    1000086,
+    1000087,
+    1000088,
+    1000089,
+    1000090,
+    1000091,
+    1000092,
+    1000093,
+    1000094,
+    1000095,
+    1000096,
+    1000097,
+    1000098,
+    1000099
+  ],
+  "issued_actions": 302228,
+  "action_counts": {
+    "1": 38075,
+    "0": 264153
+  },
+  "largest_action_fraction": 0.8740189525788478,
+  "raw_score_min": [
+    -6.0193305015563965,
+    -8.503290176391602
+  ],
+  "raw_score_max": [
+    8.566641807556152,
+    6.644220352172852
+  ],
+  "raw_score_mean": [
+    5.981686487398515,
+    -5.343010402155009
+  ]
+}

evaluations/flappy/flappy-qwenoft-hist8-scratch-s42-v2/evaluations/profile/eval_config.yaml

@@ -0,0 +1,76 @@
+experiment:
+  name: flappy-qwenoft-hist8-scratch-s42-v2-step5000-history01-profile-par32
+  seed: 1000000
+backend:
+  run_mode: eval
+executor:
+  mode: simulated
+  simulated_worker_capacity: 1
+  simulated_inference_pool: true
+  inference_devices:
+  - cuda:0
+  - cuda:1
+  inference_batch_size: 16
+env:
+  name: flappy
+  gym_id: FlappyBird-v0
+  env_fps: 10
+  obs_fps: 10
+  render_mode: rgb_array
+  observation_mode: image
+  use_lidar: false
+  normalize_obs: true
+  audio_on: false
+  frame_stack: 1
+  obs_resize:
+  - 224
+  - 224
+  simulator: gpu
+  use_gpu_render: false
+  gpu_render_device: auto
+  gpu_render_profile: false
+  gpu_render_profile_interval: 200
+  noop_action: noop
+  action_map:
+    noop: 0
+    flap: 1
+  oneshot_actions:
+  - flap
+  action_history_decisions: 8
+latency:
+  method: temporal
+  profile_path: profiles/profile.json
+  profile_worker_slot: 0
+  seed: 271828
+  add_latency_info: false
+scheduler:
+  hold_policy: one_frame_then_noop
+  ordering_policy: latest_ready
+policy:
+  type: starvla
+  checkpoint_path: checkpoints/model.pt
+  model_config_path: config.yaml
+  device: cuda:0
+  unnorm_key: new_embodiment
+  prompt_mode: latency_neutral
+  actions:
+  - noop
+  - flap
+  backbone_path: BACKBONE_SNAPSHOT
+  worker_python_executable: python
+  state_source: transport
+evaluation:
+  eval_episodes: 100
+  eval_parallel_envs: 32
+  latency_bench_env_backend: flappy_gpu_batched
+  eval_latency_values: null
+  eval_max_steps: 3600
+  eval_deterministic: true
+  eval_raw_reward: true
+logging:
+  output_dir: replay_outputs/profile
+  save_step_records: true
+  save_action_records: true
+  save_latency_records: true
+  video:
+    enabled: false

evaluations/flappy/flappy-qwenoft-hist8-scratch-s42-v2/evaluations/profile/provenance.json

@@ -0,0 +1,35 @@
+{
+  "collection": {
+    "teacher_repo": "latency-sensitive-bench/benchmark-models",
+    "teacher_revision": "af125bd372c22c5b00e326d9af1af64ea3751b56",
+    "teacher_path": "latency-aware/flappy/small-policy/sample-factory-qwenoft-rtx3090-temporal-hist8-capacity1-s42-v1",
+    "teacher_checkpoint": "best_000047872_24543232_reward_1331.159.pth",
+    "teacher_sha256": "a5fde7c3c8ac1917c9874994ad86c28d6fb0bf4d5a3d43787b447fc74b85327f",
+    "profile_path": "source_archive/hist8-vla-20260923/results/profiling/profiles/openvla/1x-rtx3090/flappy/instance_a5037b165aa0cedc/profile.json",
+    "profile_sha256": "c266cba7ebac05c7e37bc83f1f16fe4b27dab97942ff43290e6c41514f6e39fc",
+    "episodes": 200,
+    "filter_preset": "none",
+    "history": "decision_v1"
+  },
+  "training_run_id": "flappy-qwenoft-hist8-scratch-s42-v2",
+  "training_step": 5000,
+  "global_batch": 128,
+  "base_model": "Qwen/Qwen3-VL-4B-Instruct",
+  "wandb_run_id": "126a7fee",
+  "checkpoint_sha256": "c6c1f36ab709d596a9123d395377d43fa57d6e9e4b8399778f3b374ca501f883",
+  "eval_protocol_version": "hist8-history01-par32-v3",
+  "history_encoding": "binary_0_1",
+  "episodes": 100,
+  "env_condition": "profile",
+  "parallel_envs": 32,
+  "max_episode_raw_frames": 3600,
+  "inference_devices": [
+    "cuda:0",
+    "cuda:1"
+  ],
+  "episode_seeds": [
+    1000000,
+    1000099
+  ],
+  "profile_sha256": "c266cba7ebac05c7e37bc83f1f16fe4b27dab97942ff43290e6c41514f6e39fc"
+}

evaluations/flappy/flappy-qwenoft-hist8-scratch-s42-v2/evaluations/profile/resolved_config.yaml

@@ -0,0 +1,79 @@
+experiment:
+  name: flappy-qwenoft-hist8-scratch-s42-v2-step5000-history01-profile-par32
+  seed: 1000000
+backend:
+  run_mode: eval
+executor:
+  mode: simulated
+  simulated_worker_capacity: 1
+  simulated_inference_pool: true
+  inference_devices:
+  - cuda:0
+  - cuda:1
+  inference_batch_size: 16
+env:
+  name: flappy
+  gym_id: FlappyBird-v0
+  env_fps: 10
+  obs_fps: 10
+  render_mode: rgb_array
+  observation_mode: image
+  use_lidar: false
+  normalize_obs: true
+  audio_on: false
+  frame_stack: 1
+  obs_resize:
+  - 224
+  - 224
+  simulator: gpu
+  use_gpu_render: false
+  gpu_render_device: auto
+  gpu_render_profile: false
+  gpu_render_profile_interval: 200
+  noop_action: noop
+  action_map:
+    noop: 0
+    flap: 1
+  oneshot_actions:
+  - flap
+  action_history_decisions: 8
+latency:
+  method: temporal
+  profile_path: profiles/profile.json
+  profile_worker_slot: 0
+  seed: 271828
+  add_latency_info: false
+scheduler:
+  hold_policy: one_frame_then_noop
+  ordering_policy: latest_ready
+policy:
+  type: starvla
+  checkpoint_path: checkpoints/model.pt
+  model_config_path: config.yaml
+  device: cuda:0
+  unnorm_key: new_embodiment
+  prompt_mode: latency_neutral
+  actions:
+  - noop
+  - flap
+  backbone_path: BACKBONE_SNAPSHOT
+  worker_python_executable: python
+  state_source: transport
+  action_prefix:
+    mode: none
+evaluation:
+  eval_episodes: 100
+  eval_parallel_envs: 32
+  latency_bench_env_backend: flappy_gpu_batched
+  eval_latency_values: null
+  eval_max_steps: 3600
+  eval_deterministic: true
+  eval_raw_reward: true
+logging:
+  output_dir: replay_outputs/profile
+  save_step_records: true
+  save_action_records: true
+  save_latency_records: true
+  video:
+    enabled: false
+  simulated_pipeline_profile: false

evaluations/flappy/flappy-qwenoft-hist8-scratch-s42-v2/evaluations/profile/summary.json

@@ -0,0 +1,217 @@
+{
+  "checkpoint_path": "source_archive/hist8-vla-v2-20260923/outputs/tasks/decision_history_vla_hist8_v2/bundles/flappy-qwenoft-hist8-scratch-s42-v2/checkpoints/model.pt",
+  "experiment_name": "flappy-qwenoft-hist8-scratch-s42-v2-step5000-history01-profile-par32",
+  "latency": "profile_sample",
+  "latency_type": "profile_sample",
+  "lengths": [
+    3600,
+    3600,
+    3600,
+    3600,
+    3600,
+    425,
+    3600,
+    2894,
+    3600,
+    389,
+    3600,
+    3600,
+    3600,
+    3600,
+    3553,
+    3600,
+    3600,
+    3600,
+    2314,
+    3600,
+    3600,
+    3600,
+    3600,
+    3600,
+    1688,
+    3600,
+    3600,
+    163,
+    3600,
+    615,
+    2443,
+    3600,
+    2382,
+    3512,
+    3600,
+    3600,
+    2721,
+    3600,
+    3600,
+    3600,
+    3600,
+    3600,
+    3600,
+    3600,
+    466,
+    3600,
+    3177,
+    3600,
+    2197,
+    2427,
+    3371,
+    3600,
+    3589,
+    3600,
+    3600,
+    3600,
+    3600,
+    1365,
+    3600,
+    3600,
+    3600,
+    1668,
+    2953,
+    3600,
+    3600,
+    3600,
+    3600,
+    3600,
+    3600,
+    1901,
+    3600,
+    3600,
+    3596,
+    3600,
+    3600,
+    3600,
+    3600,
+    3600,
+    3600,
+    3600,
+    1673,
+    3600,
+    3177,
+    3600,
+    2016,
+    3600,
+    217,
+    3600,
+    583,
+    858,
+    1608,
+    3600,
+    620,
+    1971,
+    3600,
+    3600,
+    3600,
+    2382,
+    3600,
+    548
+  ],
+  "mean_length": 3030.62,
+  "mean_return": 373.7790043977648,
+  "returns": [
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    50.40000061690807,
+    444.6000052243471,
+    356.7000041976571,
+    444.6000052243471,
+    45.90000056475401,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    437.90000515431166,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    284.3000033572316,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    207.30000244826078,
+    444.6000052243471,
+    444.6000052243471,
+    17.900000236928463,
+    444.6000052243471,
+    73.90000089257956,
+    300.80000354349613,
+    444.6000052243471,
+    292.9000034555793,
+    432.900005094707,
+    444.6000052243471,
+    444.6000052243471,
+    334.9000039473176,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    55.40000067651272,
+    444.6000052243471,
+    391.30000460892916,
+    444.6000052243471,
+    269.9000031873584,
+    298.3000035211444,
+    415.2000048905611,
+    444.6000052243471,
+    442.4000052064657,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    166.90000198036432,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    204.40000241994858,
+    363.50000428408384,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    233.10000275820494,
+    444.6000052243471,
+    444.6000052243471,
+    443.10000521689653,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    204.90000242739916,
+    444.6000052243471,
+    391.30000460892916,
+    444.6000052243471,
+    247.30000292509794,
+    444.6000052243471,
+    25.1000003144145,
+    444.6000052243471,
+    69.80000084638596,
+    104.50000124424696,
+    197.50000233203173,
+    444.6000052243471,
+    74.40000090003014,
+    241.90000285953283,
+    444.6000052243471,
+    444.6000052243471,
+    444.6000052243471,
+    292.9000034555793,
+    444.6000052243471,
+    65.40000079572201
+  ],
+  "seed": 1000000,
+  "source_profile_path": "source_archive/hist8-vla-20260923/results/profiling/profiles/openvla/1x-rtx3090/flappy/instance_a5037b165aa0cedc/profile.json",
+  "std_return": 126.82699726383998,
+  "suite_name": "profile_sample",
+  "timestamp_utc": "2026-09-23T18:46:25.729075+00:00"
+}

evaluations/flappy/flappy-qwenoft-hist8-scratch-s42-v2/evaluations/profile/verification.json

@@ -0,0 +1,66 @@
+{
+  "job_id": "eval_v2__flappy-qwenoft-hist8-scratch-s42-v2",
+  "task": "flappy",
+  "model": "qwenoft",
+  "remote_eval_dir": "source_archive/hist8-vla-v2-20260923/outputs/tasks/decision_history_vla_hist8_v2/evals/flappy-qwenoft-hist8-scratch-s42-v2-to-profile-par32",
+  "checked_at": 1790189675.660645,
+  "training_step": 5000,
+  "checkpoint_sha256": "c6c1f36ab709d596a9123d395377d43fa57d6e9e4b8399778f3b374ca501f883",
+  "profile_sha256": "c266cba7ebac05c7e37bc83f1f16fe4b27dab97942ff43290e6c41514f6e39fc",
+  "wandb_run_id": "126a7fee",
+  "eval_protocol_version": "hist8-history01-par32-v3",
+  "max_episode_raw_frames": 3600,
+  "parallel_envs": 32,
+  "mean_return": 373.7790043977648,
+  "std_return": 126.82699726383991,
+  "mean_length": 3030.62,
+  "episodes": 100,
+  "issued_actions": 302228,
+  "top_action": "noop",
+  "largest_action_fraction": 0.8740189525788478,
+  "unique_action_values": 2,
+  "episode_final_done": 34,
+  "episodes_at_frame_cap": 66,
+  "checks": {
+    "100_exact_seeds": true,
+    "checkpoint_sha": true,
+    "bundle_step": true,
+    "profile_sha": true,
+    "binary_history": true,
+    "raw_score_means_match": true,
+    "action_counts_match": true,
+    "action_total_match": true,
+    "summary_returns_match": true,
+    "summary_lengths_match": true,
+    "100_final_steps": true,
+    "final_step_lengths_match": true,
+    "parallel_capacity_horizon": true,
+    "fps_history": true,
+    "environment_latency_matches": true,
+    "episode_latency_protocol": true,
+    "training_recipe": true,
+    "training_dimensions": true,
+    "initialization_revision": true
+  },
+  "job": {
+    "id": "eval_v2__flappy-qwenoft-hist8-scratch-s42-v2",
+    "source": "v2",
+    "python": "source_archive/home/ubuntu/miniconda3/envs/starvla_rl_games_openvla/bin/python",
+    "task": "flappy",
+    "model": "qwenoft",
+    "env_condition": "profile",
+    "parallel_envs": 32,
+    "metric_prefix": "eval_hist8_v3/profile_to_profile",
+    "kind": "eval_ready",
+    "backbone": "source_archive/hist8-vla-20260923/backbone/Qwen3-VL-4B-Instruct",
+    "profile_path": "source_archive/hist8-vla-20260923/results/profiling/profiles/openvla/1x-rtx3090/flappy/instance_a5037b165aa0cedc/profile.json",
+    "bundle_dir": "source_archive/hist8-vla-v2-20260923/outputs/tasks/decision_history_vla_hist8_v2/bundles/flappy-qwenoft-hist8-scratch-s42-v2",
+    "ready_path": "source_archive/hist8-vla-v2-20260923/outputs/tasks/decision_history_vla_hist8_v2/bundles/flappy-qwenoft-hist8-scratch-s42-v2/bundle.complete.json",
+    "eval_dir": "source_archive/hist8-vla-v2-20260923/outputs/tasks/decision_history_vla_hist8_v2/evals/flappy-qwenoft-hist8-scratch-s42-v2-to-profile-par32",
+    "gpus_required": 2,
+    "priority": 1000,
+    "depends_on": []
+  },
+  "evidence_dir": "source_archive/outputs/tasks/decision_history_vla_hist8_v2/evaluation_receipts/h100/eval_v2__flappy-qwenoft-hist8-scratch-s42-v2",
+  "verification_status": "verified"
+}