Linslab's picture
Index level-wise PI0.5 multi-task LoRA (pi05_levels_20261028): 109/109 task evaluations, 50 episodes each
529b73e verified
Raw History Blame Contribute Delete
4.82 kB
Invalid JSON:Unexpected token 'N', ..."rn_mean": NaN, "... is not valid JSON
{
"suite": "level_3",
"step": 80000,
"num_tasks": 26,
"tasks_evaluated": 26,
"evaluation_complete": true,
"pending_tasks": [],
"episodes_per_task": 50,
"episode_seeds": "1000-1049",
"mean_success_rate": 0.005384615384615385,
"per_task": {
"GroceryPrepare-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 2074.76814,
"embodiment": "rigid_body/single_arm_gripper"
},
"Stack6Blocks-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 1420.44704,
"embodiment": "rigid_body/single_arm_gripper"
},
"PushTargetPose-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": -449.60949999999997,
"embodiment": "rigid_body/single_arm_gripper"
},
"MazeBlockPush-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": -690.33772,
"embodiment": "rigid_body/single_arm_gripper"
},
"TweezersPickPlace-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": -130.1295,
"embodiment": "rigid_body/single_arm_dex"
},
"ClutterGrasp-v1": {
"success_rate": 0.04,
"successes": 2,
"return_mean": NaN,
"embodiment": "rigid_body/single_arm_dex"
},
"MobilePickPlace-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 38.18522,
"embodiment": "rigid_body/single_arm_dex"
},
"HammerNail-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": -70.39844,
"embodiment": "rigid_body/single_arm_dex"
},
"DualArmFoldCarton-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 6295.6837,
"embodiment": "rigid_body/dual_arm_gripper"
},
"DualArmOpenFridge-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 1425.1551000000002,
"embodiment": "rigid_body/dual_arm_gripper"
},
"DualArmAssembleBracket-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 25.42692,
"embodiment": "rigid_body/dual_arm_gripper"
},
"DualArmOrganizeCabinet-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 1281.50874,
"embodiment": "rigid_body/dual_arm_gripper"
},
"DualArmPingpongBalance-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 114.5293,
"embodiment": "rigid_body/dual_arm_dex"
},
"DualArmSetupTable-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 739.85418,
"embodiment": "rigid_body/dual_arm_dex"
},
"DualArmMakeLine-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 10566.17234,
"embodiment": "rigid_body/dual_arm_dex"
},
"DualArmPlugCharger-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 34.36598,
"embodiment": "rigid_body/dual_arm_dex"
},
"DualArmLayerClayInDish-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 481.77495999999996,
"embodiment": "deformable/dual_arm_dex"
},
"DualArmThreadCloth-v1": {
"success_rate": 0.04,
"successes": 2,
"return_mean": 103.5819,
"embodiment": "deformable/dual_arm_dex"
},
"DualArmThreading-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 211.09184000000002,
"embodiment": "deformable/dual_arm_dex"
},
"DualArmWeaveCloth-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 28.101260000000003,
"embodiment": "deformable/dual_arm_gripper"
},
"DualArmShapeDoll-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 410.52052000000003,
"embodiment": "deformable/dual_arm_gripper"
},
"DualArmThreadRopeThroughTube-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 1.18342,
"embodiment": "deformable/dual_arm_gripper"
},
"HitBox-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 349.82928,
"embodiment": "deformable/single_arm_dex"
},
"PressFitSoftPeg-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": 451.49809999999997,
"embodiment": "deformable/single_arm_dex"
},
"LiquidMeasuring-v1": {
"success_rate": 0.0,
"successes": 0,
"return_mean": NaN,
"embodiment": "deformable/single_arm_gripper"
},
"SeparateTwoRopesEnv-v1": {
"success_rate": 0.06,
"successes": 3,
"return_mean": 3958.67206,
"embodiment": "deformable/single_arm_gripper"
}
},
"note": "Only the final checkpoint (step 80000) was evaluated, 50 episodes per task; per-episode rows are in eval_reports/<task>.json.",
"campaign": "pi05_levels_20261028"
}