| { |
| "schema": "axis_heldout_benchmark_v2", |
| "name": "AXIS held-out benchmark", |
| "SIZE": { |
| "active_tasks": 10, |
| "conditional_tasks": 10, |
| "why": "A benchmark task must be absent from pretraining. Of 124 quality-eligible tasks only 11 are absent from the stage1_5k training ranges and the 11th (813) is unevaluable, so TEN is the ceiling against today's stage-1 models. The other ten are contaminated -- 157 episodes of the 5,000 -- and become usable only if a future stage-1 run excludes the held-out set.", |
| "active": [ |
| 809, |
| 811, |
| 866, |
| 868, |
| 945, |
| 953, |
| 966, |
| 1426, |
| 1459, |
| 1889 |
| ], |
| "conditional": [ |
| 810, |
| 815, |
| 929, |
| 970, |
| 1046, |
| 1252, |
| 1427, |
| 1458, |
| 1746, |
| 1891 |
| ] |
| }, |
| "DATA_STATUS": { |
| "state": "AWAITING RE-RENDER", |
| "camera_rig": "current demos use the superseded rig (hand->cam [-0.074,0,0.0292], fovy 51.9); the pretraining corpus uses [0.05,0,0] / fovy 74.8", |
| "appearance": "current demos are randomized per demo (TextureModder + LightingModder), so the 20 demos of a task span 20 visual worlds", |
| "request": "render_request/ in this repo covers all 20 tasks" |
| }, |
| "protocol": { |
| "adaptation_demos_per_task": 20, |
| "eval_episodes_per_task": 50, |
| "eval_requires": "--dual-sim", |
| "policy": "pi05_base + 20-demo LoRA, constant LR after 100-step warmup", |
| "gate": "keep a task if 0.20 <= success <= 0.80" |
| }, |
| "unevaluable": { |
| "813": "layout generator fails after 50,000 attempts placing 4 objects in a 0.50 m box", |
| "866": "declares DirectedRotationChecker, which is NOT in the checker registry; the harness substitutes AlwaysFalseChecker so the task returns 0/50 under any policy" |
| }, |
| "join_on": "(task_id, attempt_id) -- never episode_index", |
| "tasks": [ |
| { |
| "task_id": 809, |
| "operation": "separate", |
| "name": "separate the sandwich from the juice", |
| "status": "active", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.9229 |
| }, |
| { |
| "task_id": 810, |
| "operation": "separate", |
| "name": "separate the sandwich from the napkin box", |
| "status": "conditional", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.922 |
| }, |
| { |
| "task_id": 811, |
| "operation": "reach", |
| "name": "reach target pose", |
| "status": "active", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.932 |
| }, |
| { |
| "task_id": 815, |
| "operation": "place", |
| "name": "place with orientation", |
| "status": "conditional", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.9115 |
| }, |
| { |
| "task_id": 866, |
| "operation": "rotate", |
| "name": "rotate the computer mouse", |
| "status": "active", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.9028 |
| }, |
| { |
| "task_id": 868, |
| "operation": "pick", |
| "name": "pick single rigid object", |
| "status": "active", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.942 |
| }, |
| { |
| "task_id": 929, |
| "operation": "put_spatial", |
| "name": "put the tea cup on the left side of the tea pot", |
| "status": "conditional", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.8942 |
| }, |
| { |
| "task_id": 945, |
| "operation": "put_into", |
| "name": "put both the potato and the pepper in the bowl", |
| "status": "active", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.922 |
| }, |
| { |
| "task_id": 953, |
| "operation": "put_spatial", |
| "name": "put the muffin on the left side of the cup", |
| "status": "active", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.9151 |
| }, |
| { |
| "task_id": 966, |
| "operation": "put_on", |
| "name": "put the eyeglasses case on the tissue box", |
| "status": "active", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.9231 |
| }, |
| { |
| "task_id": 970, |
| "operation": "rotate", |
| "name": "rotate the eyeglasses case", |
| "status": "conditional", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.8546 |
| }, |
| { |
| "task_id": 1046, |
| "operation": "put_into", |
| "name": "put the toy car into the basket", |
| "status": "conditional", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.9018 |
| }, |
| { |
| "task_id": 1252, |
| "operation": "remove", |
| "name": "remove loose lid", |
| "status": "conditional", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.8896 |
| }, |
| { |
| "task_id": 1426, |
| "operation": "rearrange", |
| "name": "rearrange three items", |
| "status": "active", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.9132 |
| }, |
| { |
| "task_id": 1427, |
| "operation": "rearrange", |
| "name": "rearrange three items 2", |
| "status": "conditional", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.9097 |
| }, |
| { |
| "task_id": 1458, |
| "operation": "insert", |
| "name": "insert utensil into holder", |
| "status": "conditional", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.917 |
| }, |
| { |
| "task_id": 1459, |
| "operation": "insert", |
| "name": "fit lid onto container", |
| "status": "active", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.904 |
| }, |
| { |
| "task_id": 1746, |
| "operation": "put_on", |
| "name": "put the cube block on the poker", |
| "status": "conditional", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.914 |
| }, |
| { |
| "task_id": 1889, |
| "operation": "put_beside", |
| "name": "put the kiwi beside the banana", |
| "status": "active", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.917 |
| }, |
| { |
| "task_id": 1891, |
| "operation": "put_beside", |
| "name": "put the orange beside the banana", |
| "status": "conditional", |
| "adaptation_demos": 20, |
| "mean_reward_final": 0.9076 |
| } |
| ] |
| } |