diff --git a/README.md b/README.md index d49a6b54970e117333b01dd757ea9453d688acaf..1a1da0d40f6a58fd904f15eae7e4055be020473d 100644 --- a/README.md +++ b/README.md @@ -265,3 +265,14 @@ Each `multi_task/level_/` folder contains `lora_step_0080000.npz`, `norm_stat | DualArmLiquidMixing-v1 | 4 | | (evaluation pending) | + + +## Previous single-task checkpoints classified as reusable + +The [`previous_single_task/`](previous_single_task) collection preserves one best checkpoint from each of the 62 earlier single-task runs classified as reusable in the 2026-10-22 audit. It is separate from newer retraining runs in `single_task/`. See the [CSV](previous_single_task/manifest.csv) or [JSON manifest](previous_single_task/manifest.json), sorted by level, suite, and task. + +For 60 tasks, selection uses the available 50-episode reevaluations (seeds 1000–1049). The original **DualArmPouringWater-v1** peak is **19/20 (95%) at 30k**; the original **MazeBlockPush-v1** peak is **13/20 (65%) at 30k**. These two are selected from historical 20-episode online evaluations and must not be presented as eval50 scores. Each folder contains the selected weights, normalization statistics, evaluation summary/history, and audit provenance. + +Non-finite reward means in the historical reports are represented as null/blank and marked `return_mean_finite=false`; checkpoint selection uses the reported success counts. + +The reusable classification was based on timestamp and source-change review; the original training runs did not save complete source snapshots. diff --git a/previous_single_task/BuildTent-v1/eval_result.json b/previous_single_task/BuildTent-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..81fe33068694939b023d12b44c220d6f7bc1734d --- /dev/null +++ b/previous_single_task/BuildTent-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "BuildTent-v1", + "level": "level_4", + "suite": "deformable/single_arm_gripper", + "step": 10000, + "episodes": 50, + "successes": 3, + "success_rate": 0.06, + "return_mean": 0.3800000047683716, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/BuildTent-v1/evaluation_history.csv b/previous_single_task/BuildTent-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..9e533165320891306b4b94362b55c98a15b7013e --- /dev/null +++ b/previous_single_task/BuildTent-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,2,0.04,0.3500000035762787,True +10000,50,3,0.06,0.3800000047683716,True +15000,50,0,0.0,0.17800000190734863,True +20000,50,2,0.04,0.33200000286102294,True +25000,50,1,0.02,0.26600000381469724,True +30000,50,1,0.02,0.21800000071525574,True +35000,50,2,0.04,0.36600000500679014,True +40000,50,1,0.02,0.25800000309944154,True diff --git a/previous_single_task/BuildTent-v1/lora_step_0010000.npz b/previous_single_task/BuildTent-v1/lora_step_0010000.npz new file mode 100644 index 0000000000000000000000000000000000000000..069aaa775298d2786c941b1b44bc89f4af0ada8c --- /dev/null +++ b/previous_single_task/BuildTent-v1/lora_step_0010000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:d27916914df184fddd0df81bde6d7f7187f76d203648989b74d2711fca67281d +size 1867844866 diff --git a/previous_single_task/BuildTent-v1/metadata.json b/previous_single_task/BuildTent-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..d38a365d781bf4fed64910b05a378b2516beb6ab --- /dev/null +++ b/previous_single_task/BuildTent-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "BuildTent-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0010000.npz", + "checkpoint_sha256": "d27916914df184fddd0df81bde6d7f7187f76d203648989b74d2711fca67281d", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 10000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "8a808e9424a3b654d90a7bf7a460c705d550a33f93f82b2330f6e6d1664e9470", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_BuildTent-v1", + "source_path": "maniladder/envs/deformable/single_arm_gripper/level_4/build_tent.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/BuildTent-v1/norm_stats.json b/previous_single_task/BuildTent-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..653e84975d55e2bd2c27cd960ca26f50aaafc4a4 --- /dev/null +++ b/previous_single_task/BuildTent-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0025500021874904633, + -0.0027968876529484987, + -0.012523707933723927, + 0.0001537687494419515, + -0.0020131554920226336, + 0.001333035295829177, + -0.20729388296604156, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.04087639972567558, + 0.03963173180818558, + 0.04210036247968674, + 0.00272456300444901, + 0.005590345244854689, + 0.045177798718214035, + 0.6698947548866272, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.07999999821186066, + -0.07999999821186066, + -0.07999999821186066, + -0.008506598155200482, + -0.0363615726865828, + -0.11999999731779099, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.07996799821257591, + 0.07996799821257591, + 0.07996799821257591, + 0.00948646602630615, + 0.0019109909217804658, + 0.11995199731886386, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/BuildTent-v1/source_audit.json b/previous_single_task/BuildTent-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..39c800e2f3a42b8c8bbe2e0d7948941dcabfdb15 --- /dev/null +++ b/previous_single_task/BuildTent-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "BuildTent-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.1", + "peak_step": "5000", + "final": "0.0", + "start": "2026-09-04 06:24:15", + "eval_mtime": "2026-09-07 12:22:10.813013", + "source_mtime": "2026-08-29 21:10:09.959756", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/single_arm_gripper/level_4/build_tent.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_BuildTent-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_BuildTent-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_BuildTent-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/ClothWipe-v1/eval_result.json b/previous_single_task/ClothWipe-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..b9defb87958f5b5bb56d32793b150f34761b9a47 --- /dev/null +++ b/previous_single_task/ClothWipe-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "ClothWipe-v1", + "level": "level_2", + "suite": "deformable/single_arm_dex", + "step": 25000, + "episodes": 50, + "successes": 26, + "success_rate": 0.52, + "return_mean": 354.6896101513505, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/ClothWipe-v1/evaluation_history.csv b/previous_single_task/ClothWipe-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..3f8d582974f993051201330a53928872ce37aabf --- /dev/null +++ b/previous_single_task/ClothWipe-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,8,0.16,366.58430208094416,True +10000,50,15,0.3,373.01743837943303,True +15000,50,10,0.2,384.1779712532461,True +20000,50,19,0.38,363.6087702172995,True +25000,50,26,0.52,354.6896101513505,True +30000,50,17,0.34,383.1513956150413,True +35000,50,24,0.48,343.40371913433074,True +40000,50,11,0.22,391.8261306717247,True diff --git a/previous_single_task/ClothWipe-v1/lora_step_0025000.npz b/previous_single_task/ClothWipe-v1/lora_step_0025000.npz new file mode 100644 index 0000000000000000000000000000000000000000..b514995a44732da4fa2545ae485e172d6de044ea --- /dev/null +++ b/previous_single_task/ClothWipe-v1/lora_step_0025000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1603d8b01f721d5ea488ada553dc31fcfad4ac54785d00a639f3d9243e13ae26 +size 1867844866 diff --git a/previous_single_task/ClothWipe-v1/metadata.json b/previous_single_task/ClothWipe-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..63bbc3490b8b11043798822df3628ee322da10f6 --- /dev/null +++ b/previous_single_task/ClothWipe-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "ClothWipe-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0025000.npz", + "checkpoint_sha256": "1603d8b01f721d5ea488ada553dc31fcfad4ac54785d00a639f3d9243e13ae26", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 25000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "3737a4befc869fded689f27e42f42c2c08ad6bc4d55aa6d7926caebb26b0ef96", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_ClothWipe-v1", + "source_path": "maniladder/envs/deformable/single_arm_dex/level_2/cloth_wipe.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/ClothWipe-v1/norm_stats.json b/previous_single_task/ClothWipe-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..0c933d0e36602fb5d09b54f2a083e17c98161c1c --- /dev/null +++ b/previous_single_task/ClothWipe-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.021840250119566917, + 0.01377252396196127, + -0.011951087042689323, + 0.0, + 0.0, + 0.0, + -0.0020442188251763582, + 0.04531954973936081, + 0.031206944957375526, + 0.01766965724527836, + 0.00003805236337939277, + 0.010735918767750263, + 0.005258867982774973, + 0.004407794680446386, + 0.00017677324649412185, + 0.007383469957858324, + 0.0058954618871212006, + 0.00477808341383934, + 0.024601222947239876, + -0.04246635362505913, + 0.09596503525972366, + 0.04429228976368904, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.026049785315990448, + 0.05998889356851578, + 0.04185228794813156, + 0.0, + 0.0, + 0.0, + 0.002836348256096244, + 0.11692291498184204, + 0.10987512022256851, + 0.1109260767698288, + 0.0006267151911742985, + 0.1152331605553627, + 0.11674018949270248, + 0.11585366725921631, + 0.0007009835680946708, + 0.119076207280159, + 0.11743780970573425, + 0.11624316871166229, + 0.15012766420841217, + 0.04029984399676323, + 0.17632104456424713, + 0.16012516617774963, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.0052800000786781265, + -0.20000000298023224, + -0.1000268955245614, + 0.0, + 0.0, + 0.0, + -0.008603449368476868, + -0.28609981751441954, + -0.2901937688648701, + -0.31563101294636725, + -0.002254950039461255, + -0.33100634697675707, + -0.34734427826404574, + -0.3471908022046089, + -0.0008490206276997925, + -0.34733089823126795, + -0.347342262673378, + -0.34718933740854263, + -0.20072315499782561, + -0.09652527822162738, + -0.3606807085394859, + -0.4234354164123535, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.10000000149011612, + 0.09994000148922205, + 0.09995552141517403, + 0.0, + 0.0, + 0.0, + 0.002720086799934506, + 0.34715260868072506, + 0.34708210074305534, + 0.34708095822334295, + 0.002635717754811049, + 0.3471322993397712, + 0.3471015685796738, + 0.34710380837321275, + 0.0037161918932572006, + 0.3470775063335896, + 0.34710146129131325, + 0.3471038936078549, + 0.7475124543190003, + -0.000013326572156441285, + 0.4806724551320075, + 0.4806724341869354, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/ClothWipe-v1/source_audit.json b/previous_single_task/ClothWipe-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..0d1b7b7a7c86a08f76c158c19eff1aa58b3faadf --- /dev/null +++ b/previous_single_task/ClothWipe-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "ClothWipe-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.35", + "peak_step": "10000", + "final": "0.2", + "start": "2026-09-04 06:14:15", + "eval_mtime": "2026-09-07 18:53:31.728347", + "source_mtime": "2026-08-22 02:28:27.164133", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/single_arm_dex/level_2/cloth_wipe.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_ClothWipe-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_ClothWipe-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_ClothWipe-v1/lora_params/lora_step_0010000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmBallClassify-v1/eval_result.json b/previous_single_task/DualArmBallClassify-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..9061ec7670f89f791f7cf1407f4c259258ad4f1c --- /dev/null +++ b/previous_single_task/DualArmBallClassify-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmBallClassify-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_gripper", + "step": 10000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": 3305.200233145058, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 7, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmBallClassify-v1/evaluation_history.csv b/previous_single_task/DualArmBallClassify-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..a87050020d955aaa308a4aa5e5c79ee1f1f130b2 --- /dev/null +++ b/previous_single_task/DualArmBallClassify-v1/evaluation_history.csv @@ -0,0 +1,8 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +10000,50,0,0.0,3305.200233145058,True +15000,50,0,0.0,2958.5908711349966,True +20000,50,0,0.0,2099.2708752262592,True +25000,50,0,0.0,2537.1459140840175,True +30000,50,0,0.0,2231.6329792255165,True +35000,50,0,0.0,2274.1544187426566,True +40000,50,0,0.0,2516.529187646508,True diff --git a/previous_single_task/DualArmBallClassify-v1/lora_step_0010000.npz b/previous_single_task/DualArmBallClassify-v1/lora_step_0010000.npz new file mode 100644 index 0000000000000000000000000000000000000000..850253582053c959dbbea7e390e0c27e5df04378 --- /dev/null +++ b/previous_single_task/DualArmBallClassify-v1/lora_step_0010000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0bec5644e89c77f1fbdb520eae5ea4504903d0bbed7dab3b41295d045e50aa0f +size 1867844866 diff --git a/previous_single_task/DualArmBallClassify-v1/metadata.json b/previous_single_task/DualArmBallClassify-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..cba28da391cc13c108c76d0820d03161de619985 --- /dev/null +++ b/previous_single_task/DualArmBallClassify-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmBallClassify-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0010000.npz", + "checkpoint_sha256": "0bec5644e89c77f1fbdb520eae5ea4504903d0bbed7dab3b41295d045e50aa0f", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 10000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "c01e6c52713848e26c0730d8ae721a0afd621438befcc251a87d6d7783bd3216", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmBallClassify-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_gripper/level_4/ball_classify.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmBallClassify-v1/norm_stats.json b/previous_single_task/DualArmBallClassify-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..fe6446a49928f73a3ad9aa84351f28c5141b40a8 --- /dev/null +++ b/previous_single_task/DualArmBallClassify-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.003652821993455291, + 0.00009933079854818061, + -0.0033298791386187077, + -0.0006109398091211915, + 0.00002821255475282669, + -0.00019378938304726034, + 0.5015915036201477, + -0.003281553275883198, + -0.00003442637171247043, + -0.002978231757879257, + 0.0005717254243791103, + -0.000028130618375143968, + -0.00010722584556788206, + 0.5014796853065491, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.04649297520518303, + 0.03498956933617592, + 0.04953033849596977, + 0.0038567353039979935, + 0.0008663812186568975, + 0.0762215405702591, + 0.8651045560836792, + 0.046709612011909485, + 0.03504866361618042, + 0.049159467220306396, + 0.003638111287727952, + 0.0007019665790721774, + 0.07396870106458664, + 0.8651694059371948, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + -0.023631684099137785, + -0.001138090783730148, + -0.057200028097629585, + -1.0, + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + -0.0011596357149071986, + -0.0021176266670227054, + -0.04959999999999998, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.001178614109754561, + 0.0021125587821006764, + 0.05279996862411496, + 0.9996, + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0213701859745197, + 0.000994965881109238, + 0.04520000000000013, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmBallClassify-v1/source_audit.json b/previous_single_task/DualArmBallClassify-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..f2e6251bf9e94bf85438bf4e1ec616d8b62ea360 --- /dev/null +++ b/previous_single_task/DualArmBallClassify-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmBallClassify-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "10000", + "final": "0.0", + "start": "2026-08-22 07:31:06", + "eval_mtime": "2026-08-26 02:55:26.611068", + "source_mtime": "2026-07-31 14:04:42.304658", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_gripper/level_4/ball_classify.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmBallClassify-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmBallClassify-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmBallClassify-v1/lora_params/lora_step_0010000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmBilliardBridge-v1/eval_result.json b/previous_single_task/DualArmBilliardBridge-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..19825de040aca34b47a26d9de26688cafb7c1732 --- /dev/null +++ b/previous_single_task/DualArmBilliardBridge-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmBilliardBridge-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": 1187.845783996582, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmBilliardBridge-v1/evaluation_history.csv b/previous_single_task/DualArmBilliardBridge-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..3510aed3663e6db8079ecc3e6b42228bcdc87885 --- /dev/null +++ b/previous_single_task/DualArmBilliardBridge-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,1187.845783996582,True +10000,50,0,0.0,1188.226076245308,True +15000,50,0,0.0,1175.9384981024266,True +20000,50,0,0.0,1180.3857240962982,True +25000,50,0,0.0,1189.2712672090531,True +30000,50,0,0.0,1176.501679096222,True +35000,50,0,0.0,1173.2626583218575,True +40000,50,0,0.0,1188.2963494682313,True diff --git a/previous_single_task/DualArmBilliardBridge-v1/lora_step_0005000.npz b/previous_single_task/DualArmBilliardBridge-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..f54c1e10218dc1d77cbcb26c83cb8269d8555a87 --- /dev/null +++ b/previous_single_task/DualArmBilliardBridge-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7dabff95c821665a5f6ad6d381ca64ce8dc8ae1baa11aa6b1cc042d3ebc6bc9e +size 1867844866 diff --git a/previous_single_task/DualArmBilliardBridge-v1/metadata.json b/previous_single_task/DualArmBilliardBridge-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..36c6e97d06ff3fa39d9bf23bceea23419315340e --- /dev/null +++ b/previous_single_task/DualArmBilliardBridge-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmBilliardBridge-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "7dabff95c821665a5f6ad6d381ca64ce8dc8ae1baa11aa6b1cc042d3ebc6bc9e", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "69f00f5b3727c916a815169f70c96af4a47be444ae14971ef1bfc4be01c8ed2b", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmBilliardBridge-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_gripper/level_4/hit_billiard_bridge.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmBilliardBridge-v1/norm_stats.json b/previous_single_task/DualArmBilliardBridge-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..4164dbdc4c6b5183941866927232f989e2baa2d7 --- /dev/null +++ b/previous_single_task/DualArmBilliardBridge-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.443775475025177, + -1.0803121328353882, + 0.9168494939804077, + 0.005321680102497339, + 0.009217350743710995, + 0.005060478579252958, + 0.00006726973515469581, + 0.00016356103878933936, + 0.00397635530680418, + 0.07310269773006439, + 0.19110512733459473, + 0.4165840744972229, + 0.000019972432710346766, + 0.0034384315367788076, + -0.00006220188515726477, + -0.008566319942474365, + 0.0, + 0.0, + 0.0, + 1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.4176228940486908, + 0.41765546798706055, + 0.758634626865387, + 0.026743857190012932, + 0.05443078652024269, + 0.03999131917953491, + 0.0014920177636668086, + 0.0018249147105962038, + 0.04245031625032425, + 0.9973244071006775, + 0.10192536562681198, + 0.044475995004177094, + 0.00005243471605353989, + 0.016967790201306343, + 0.00024510783259756863, + 0.02831079065799713, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.1499999761581421, + -1.6061335802078247, + -0.0006327221635729074, + -0.06952475245296955, + -0.003284824514389038, + -0.11999999731779099, + -0.0016209461279213452, + -0.0005706165596842767, + -0.021572759443521494, + -1.0, + 0.0, + 0.38027092938423157, + -0.00005133662756998092, + 0.0, + -0.0011845186471939088, + -0.11999999731779099, + 0.0, + 0.0, + 0.0, + 1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 1.093981245326996, + -0.4425879517555238, + 1.580170397590613, + 0.11995599342882632, + 0.3881123305797577, + 0.11995199731886386, + 0.0058214208092540465, + 0.005747826667129995, + 0.349866766089201, + 0.9996, + 0.24995, + 0.499975967669487, + 0.00018247468594927342, + 0.11997599731832742, + -3.5512447357343574e-8, + -0.000023999999463564237, + 0.0, + 0.0, + 0.0, + 1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmBilliardBridge-v1/source_audit.json b/previous_single_task/DualArmBilliardBridge-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..c6081948c3f6651ebb10afe44a7be43b903e2789 --- /dev/null +++ b/previous_single_task/DualArmBilliardBridge-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmBilliardBridge-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 05:48:56", + "eval_mtime": "2026-08-26 00:58:58.033045", + "source_mtime": "2026-05-14 07:01:36.053461", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_gripper/level_4/hit_billiard_bridge.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmBilliardBridge-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmBilliardBridge-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmBilliardBridge-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmButtonPress-v1/eval_result.json b/previous_single_task/DualArmButtonPress-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..eea2bac13670863dd5133c47457ce044eb644cd5 --- /dev/null +++ b/previous_single_task/DualArmButtonPress-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmButtonPress-v1", + "level": "level_1", + "suite": "rigid_body/dual_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 49, + "success_rate": 0.98, + "return_mean": 97.12502707958221, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmButtonPress-v1/evaluation_history.csv b/previous_single_task/DualArmButtonPress-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..3b619e1c324ca8a1fcc2de6afb88f23e98934920 --- /dev/null +++ b/previous_single_task/DualArmButtonPress-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,49,0.98,97.12502707958221,True +10000,50,47,0.94,105.99478406906128,True +15000,50,46,0.92,110.15575822591782,True +20000,50,43,0.86,124.01788116931915,True +25000,50,43,0.86,123.27124275445938,True +30000,50,42,0.84,128.5457330584526,True +35000,50,44,0.88,118.79112421274185,True +40000,50,43,0.86,122.83211128234863,True diff --git a/previous_single_task/DualArmButtonPress-v1/lora_step_0005000.npz b/previous_single_task/DualArmButtonPress-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..7ce43e72fbbc99310fe5261a6aa4022eacf7d2b6 --- /dev/null +++ b/previous_single_task/DualArmButtonPress-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fe6b5128698f903f0b7a412f09be569043fdb7f2ae8cb8eb2edc016ff77b1111 +size 1867844866 diff --git a/previous_single_task/DualArmButtonPress-v1/metadata.json b/previous_single_task/DualArmButtonPress-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..08197a92c8cf7811a9710d0a47977bcce226626b --- /dev/null +++ b/previous_single_task/DualArmButtonPress-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmButtonPress-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "fe6b5128698f903f0b7a412f09be569043fdb7f2ae8cb8eb2edc016ff77b1111", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "69f4cdb9de6e1d2cf50c3c8151bfdbce3cc4bf8894e1e838ddcf3a12746f4791", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmButtonPress-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_gripper/level_1/button_press.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmButtonPress-v1/norm_stats.json b/previous_single_task/DualArmButtonPress-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..e825a89c508cb35b2c3d32436bbdba621e57e0fe --- /dev/null +++ b/previous_single_task/DualArmButtonPress-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0056900507770478725, + -0.009062274359166622, + -0.09967769682407379, + 0.0, + 0.0, + 0.0, + -1.0, + 0.004136399365961552, + -0.00029441056540235877, + -0.09963323175907135, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.05193447321653366, + 0.04080270230770111, + 0.0031263765413314104, + 0.0, + 0.0, + 0.0, + 0.0, + 0.05348271131515503, + 0.019102849066257477, + 0.003366906428709626, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.04311606287956238, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -1.0, + -0.04462268203496933, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09997137827724217, + 0.02568356383405626, + -0.053108297288417816, + 0.0, + 0.0, + 0.0, + -1.0, + 0.09997107695341109, + 0.015520000231266026, + -0.04789870232343674, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmButtonPress-v1/source_audit.json b/previous_single_task/DualArmButtonPress-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..354fbd8d3327886afad84523ebb4bc459650aab3 --- /dev/null +++ b/previous_single_task/DualArmButtonPress-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmButtonPress-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.95", + "peak_step": "15000", + "final": "0.7", + "start": "2026-08-21 13:36:43", + "eval_mtime": "2026-08-24 11:02:47.845654", + "source_mtime": "2026-05-22 16:10:36.799582", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_gripper/level_1/button_press.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmButtonPress-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmButtonPress-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmButtonPress-v1/lora_params/lora_step_0015000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmClothAlign-v1/eval_result.json b/previous_single_task/DualArmClothAlign-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..3332303fb6e029568b2ee2235d8fd999f08b7e07 --- /dev/null +++ b/previous_single_task/DualArmClothAlign-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmClothAlign-v1", + "level": "level_2", + "suite": "deformable/dual_arm_gripper", + "step": 40000, + "episodes": 50, + "successes": 19, + "success_rate": 0.38, + "return_mean": 30.897125910464673, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmClothAlign-v1/evaluation_history.csv b/previous_single_task/DualArmClothAlign-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..d8fc782d619135dd9863a7928766c46902bbb7cf --- /dev/null +++ b/previous_single_task/DualArmClothAlign-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,11,0.22,33.58352347676642,True +10000,50,17,0.34,32.044271623790266,True +15000,50,13,0.26,36.29374010687694,True +20000,50,11,0.22,34.49777565704659,True +25000,50,18,0.36,29.255039263889195,True +30000,50,8,0.16,39.45685027938336,True +35000,50,17,0.34,31.9937991212029,True +40000,50,19,0.38,30.897125910464673,True diff --git a/previous_single_task/DualArmClothAlign-v1/lora_step_0040000.npz b/previous_single_task/DualArmClothAlign-v1/lora_step_0040000.npz new file mode 100644 index 0000000000000000000000000000000000000000..06adecf356c3694978bd17cfc52af85e9379fdc7 --- /dev/null +++ b/previous_single_task/DualArmClothAlign-v1/lora_step_0040000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c17241b647ad68f11aacc420f950dc5333abd47e7e665e9c0dc84af567943bac +size 1867844866 diff --git a/previous_single_task/DualArmClothAlign-v1/metadata.json b/previous_single_task/DualArmClothAlign-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..c3ad473a0a35a65b7ed54b1d5f05e3b826595863 --- /dev/null +++ b/previous_single_task/DualArmClothAlign-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmClothAlign-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0040000.npz", + "checkpoint_sha256": "c17241b647ad68f11aacc420f950dc5333abd47e7e665e9c0dc84af567943bac", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 40000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "5644e45431210009fed0603c95da2b71dd52b2d5baa4fcd6d7819517310eb94c", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmClothAlign-v1", + "source_path": "maniladder/envs/deformable/dual_arm_gripper/level_2/cloth_align.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmClothAlign-v1/norm_stats.json b/previous_single_task/DualArmClothAlign-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..28a7458ddc70b67f9488a404958e5bbce17d7cf8 --- /dev/null +++ b/previous_single_task/DualArmClothAlign-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.033096008002758026, + 0.011338340118527412, + -0.002473824890330434, + 0.0, + 0.0, + 0.0, + 0.0897277295589447, + 0.03864385932683945, + -0.007259005215018988, + -0.0022094096057116985, + 0.0, + 0.0, + 0.0, + 0.0897277295589447, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.08023522049188614, + 0.06315421313047409, + 0.09576491266489029, + 0.0, + 0.0, + 0.0, + 0.995966374874115, + 0.07438897341489792, + 0.06507132947444916, + 0.09571585059165955, + 0.0, + 0.0, + 0.0, + 0.995966374874115, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.15000000596046448, + -0.15012000596523284, + -0.15000000596046448, + 0.0, + 0.0, + 0.0, + -1.0, + -0.036084001433849336, + -0.11282400448322297, + -0.15000000596046448, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.17993400714993474, + 0.1799280071496963, + 0.17993400714993474, + 0.0, + 0.0, + 0.0, + 1.0, + 0.17993400714993474, + 0.1799280071496963, + 0.17993400714993474, + 0.0, + 0.0, + 0.0, + 1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmClothAlign-v1/source_audit.json b/previous_single_task/DualArmClothAlign-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..ecf2ba982d2b01bc74798e7c7a72df53f2bea481 --- /dev/null +++ b/previous_single_task/DualArmClothAlign-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmClothAlign-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.4", + "peak_step": "25000", + "final": "0.35", + "start": "2026-08-21 13:36:45", + "eval_mtime": "2026-08-24 13:18:20.199611", + "source_mtime": "2026-05-14 07:01:36.005461", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_gripper/level_2/cloth_align.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmClothAlign-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmClothAlign-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmClothAlign-v1/lora_params/lora_step_0025000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmClothHanger-v1/eval_result.json b/previous_single_task/DualArmClothHanger-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..b7757638476a42f0ce0c4b08d98c0c61178ea449 --- /dev/null +++ b/previous_single_task/DualArmClothHanger-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmClothHanger-v1", + "level": "level_4", + "suite": "deformable/dual_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": 144.01372902154924, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmClothHanger-v1/evaluation_history.csv b/previous_single_task/DualArmClothHanger-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..fbd99e2685e56d9f6b9469bbcd66d67228f7d5dd --- /dev/null +++ b/previous_single_task/DualArmClothHanger-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,144.01372902154924,True +10000,50,0,0.0,111.90421493053437,True +15000,50,0,0.0,126.82437116622924,True +20000,50,0,0.0,159.40660739541053,True +25000,50,0,0.0,145.93114191770553,True +30000,50,0,0.0,121.72392417907714,True +35000,50,0,0.0,118.61773026943207,True +40000,50,0,0.0,116.59107458353043,True diff --git a/previous_single_task/DualArmClothHanger-v1/lora_step_0005000.npz b/previous_single_task/DualArmClothHanger-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..a8fa74ab9cd9b42d1d327c5fd9d02d3c0859254d --- /dev/null +++ b/previous_single_task/DualArmClothHanger-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7c5dae817921bdcdf2474ea67dc5e7edd59fb41832db3e06af5ffc9f1f60c28c +size 1867844866 diff --git a/previous_single_task/DualArmClothHanger-v1/metadata.json b/previous_single_task/DualArmClothHanger-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..d549e02e79969952b61113849cd8c536bd3162b1 --- /dev/null +++ b/previous_single_task/DualArmClothHanger-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmClothHanger-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "7c5dae817921bdcdf2474ea67dc5e7edd59fb41832db3e06af5ffc9f1f60c28c", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "23c0462e0fa6d486730237be41d06a4b54a96d71283e57189fecb6dcb377de33", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmClothHanger-v1", + "source_path": "maniladder/envs/deformable/dual_arm_gripper/level_4/cloth_hanger.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmClothHanger-v1/norm_stats.json b/previous_single_task/DualArmClothHanger-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..37caeaf8d13403c97d07423e5fed7d7b777dd5f1 --- /dev/null +++ b/previous_single_task/DualArmClothHanger-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.004077791702002287, + -0.00007221452688099816, + 0.0009116464643739164, + -0.000017553806173964404, + 0.00011853600881295279, + -0.000019489245460135862, + 0.0617559477686882, + -0.0007550496375188231, + -0.00003319975076010451, + -0.00007243158324854448, + 0.0, + 0.0, + 0.0, + 0.6171345114707947, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.048190873116254807, + 0.11470469832420349, + 0.034877240657806396, + 0.02383163385093212, + 0.002447220031172037, + 0.003753161756321788, + 0.9980912804603577, + 0.03378121182322502, + 0.117764413356781, + 0.04762253910303116, + 0.0, + 0.0, + 0.0, + 0.7868577241897583, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.06009600516557689, + -0.550000011920929, + -0.05999999865889549, + -0.10000000149011612, + -0.0001756865441799177, + -0.0012627975970506689, + -1.0, + -0.08029047811031342, + -0.550000011920929, + -0.05999999865889549, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.11986599731594327, + 0.5497800119161607, + 0.11999358618780973, + 0.09996000148952008, + 0.0024361121118068665, + 0.0026999291956424643, + 0.9996, + 0.07970633983612063, + 0.5497800119161607, + 0.23865600652098656, + 0.0, + 0.0, + 0.0, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmClothHanger-v1/source_audit.json b/previous_single_task/DualArmClothHanger-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..0cc707d92027ed9bce09e673872ea76877c1215e --- /dev/null +++ b/previous_single_task/DualArmClothHanger-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmClothHanger-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-09-04 06:34:15", + "eval_mtime": "2026-09-08 01:23:42.396958", + "source_mtime": "2026-09-03 12:31:48.420123", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_gripper/level_4/cloth_hanger.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmClothHanger-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmClothHanger-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmClothHanger-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmCoverCloth-v1/eval_result.json b/previous_single_task/DualArmCoverCloth-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..7a26855ec8e00df4f4f2cb9320a0fc8f5cb644bd --- /dev/null +++ b/previous_single_task/DualArmCoverCloth-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmCoverCloth-v1", + "level": "level_1", + "suite": "deformable/dual_arm_dex", + "step": 15000, + "episodes": 50, + "successes": 40, + "success_rate": 0.8, + "return_mean": 144.4165011787426, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmCoverCloth-v1/evaluation_history.csv b/previous_single_task/DualArmCoverCloth-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..6cb7de5c46e0e8d24b6f32175dd1a0639231cf71 --- /dev/null +++ b/previous_single_task/DualArmCoverCloth-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,12,0.24,192.1267825967567,True +10000,50,17,0.34,198.25729583896697,True +15000,50,40,0.8,144.4165011787426,True +20000,50,20,0.4,181.82558477331418,True +25000,50,11,0.22,193.93926194417057,True +30000,50,25,0.5,162.4261789018489,True +35000,50,11,0.22,184.13170099303127,True +40000,50,21,0.42,162.59602199232438,True diff --git a/previous_single_task/DualArmCoverCloth-v1/lora_step_0015000.npz b/previous_single_task/DualArmCoverCloth-v1/lora_step_0015000.npz new file mode 100644 index 0000000000000000000000000000000000000000..6d9892eab39c64f4a8da7c807c2d73780b19553a --- /dev/null +++ b/previous_single_task/DualArmCoverCloth-v1/lora_step_0015000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:05fe236e482b4a9e6046ac1448bfe952538ed4b7eb217a0961a0e5228f4c91f4 +size 1867943218 diff --git a/previous_single_task/DualArmCoverCloth-v1/metadata.json b/previous_single_task/DualArmCoverCloth-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..3eb0989a17cf98748b0c1198d2338be03370e7fb --- /dev/null +++ b/previous_single_task/DualArmCoverCloth-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmCoverCloth-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0015000.npz", + "checkpoint_sha256": "05fe236e482b4a9e6046ac1448bfe952538ed4b7eb217a0961a0e5228f4c91f4", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 15000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "950a8bef9cee4af4348f7c01db34c81bb67393f6681c1fca93a468d0f392777d", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmCoverCloth-v1", + "source_path": "maniladder/envs/deformable/dual_arm_dex/level_1/cloth_cover.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmCoverCloth-v1/norm_stats.json b/previous_single_task/DualArmCoverCloth-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..ca74bbb66de6c3e662add5e35e05db8594318fd0 --- /dev/null +++ b/previous_single_task/DualArmCoverCloth-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.020830711349844933, + 0.005153999663889408, + -0.003726587165147066, + 0.0, + 0.0, + 0.0, + 0.0007814383134245872, + 0.03908273205161095, + 0.01537161972373724, + 0.008488941006362438, + 2.949460338186327e-7, + 0.002083374420180917, + 0.0016338457353413105, + 0.0014616945991292596, + 0.00009624911763239652, + 0.0020686346106231213, + 0.0015216302126646042, + 0.001409596879966557, + 0.013187290169298649, + -0.030535440891981125, + 0.09771828353404999, + 0.033453188836574554, + 0.03169100359082222, + -0.0067009879276156425, + -0.0029890481382608414, + 0.0, + 0.0, + 0.0, + 0.0005027826991863549, + 0.04033196344971657, + 0.013476350344717503, + 0.005203113425523043, + -5.334342404239578e-6, + 0.0016126006375998259, + 0.0014715794241055846, + 0.0013835248537361622, + -7.543630090367515e-6, + 0.002036484656855464, + 0.001698843901976943, + 0.0014837796334177256, + -0.018387144431471825, + 0.026202918961644173, + 0.10715342313051224, + 0.03413424268364906 + ], + "std": [ + 0.07190023362636566, + 0.03571072220802307, + 0.06727664172649384, + 0.0, + 0.0, + 0.0, + 0.0019879464525729418, + 0.12679819762706757, + 0.12359368801116943, + 0.12790723145008087, + 0.000017768043107935227, + 0.131780207157135, + 0.1319013237953186, + 0.13177335262298584, + 0.0004601175896823406, + 0.13216738402843475, + 0.13182371854782104, + 0.1317351907491684, + 0.11976521462202072, + 0.032983794808387756, + 0.192195862531662, + 0.18070974946022034, + 0.0628955066204071, + 0.03439631313085556, + 0.06753231585025787, + 0.0, + 0.0, + 0.0, + 0.003342552576214075, + 0.12573549151420593, + 0.12500622868537903, + 0.1307806670665741, + 0.000030190520192263648, + 0.131858691573143, + 0.13177144527435303, + 0.13171105086803436, + 0.00006363788997987285, + 0.13219822943210602, + 0.13194811344146729, + 0.13178779184818268, + 0.11761917918920517, + 0.02818627469241619, + 0.18671388924121857, + 0.17969433963298798 + ], + "q01": [ + -0.11500000208616257, + -0.14499999582767487, + -0.07503061838150024, + 0.0, + 0.0, + 0.0, + -0.0024356001280248164, + -0.31337204355001447, + -0.34634883692264556, + -0.37798575248718264, + -0.00006560978291090578, + -0.3994157726049423, + -0.4019018824994564, + -0.40175270125865936, + -0.000200373912230134, + -0.4017059543609619, + -0.40176562908887864, + -0.4017626451730728, + -0.26373545019626615, + -0.10637218671850861, + -0.3863071702957153, + -0.4992640000104904, + -0.11500000208616257, + -0.14499999582767487, + -0.07500754480361939, + 0.0, + 0.0, + 0.0, + -0.007927665764465928, + -0.3085745283961296, + -0.35420858836174013, + -0.3865451649904251, + -0.00017332696293015034, + -0.401906220972538, + -0.4019155925512314, + -0.4017590447902679, + -0.0002702471147524193, + -0.40173949862718583, + -0.4017515667676926, + -0.4017581875562668, + -0.41999998688697815, + -0.001903267851471901, + -0.339193333619833, + -0.4996320000052452 + ], + "q99": [ + 0.1449479958280921, + 0.07499399784207345, + 0.14495361500680445, + 0.0, + 0.0, + 0.0, + 0.006233786537498235, + 0.3520549351453781, + 0.3518785512566567, + 0.35260462508201595, + 0.00006584885778138414, + 0.35184360020160677, + 0.3521516149461269, + 0.3526404103517533, + 0.0020510890870355067, + 0.3521327980041504, + 0.35231300079822536, + 0.3526401241064072, + 0.4198576606094837, + 0.0024210556428879426, + 0.4198317257940769, + 0.41981598688960076, + 0.1449479958280921, + 0.11495599669218065, + 0.14495361987054348, + 0.0, + 0.0, + 0.0, + 0.006029593871906401, + 0.3519273076891899, + 0.3518693425655365, + 0.3524404733180999, + 9.003743296489078e-6, + 0.3521313259720802, + 0.3521514142751694, + 0.35264022774696346, + 0.0001877113755326717, + 0.35247879608273514, + 0.35247552341222765, + 0.35264025242328645, + 0.2425865155160427, + 0.0883572698354721, + 0.41983780074715615, + 0.41981598688960076 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmCoverCloth-v1/source_audit.json b/previous_single_task/DualArmCoverCloth-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..dfe0088641b076635eb84970b42b433b3e7f5621 --- /dev/null +++ b/previous_single_task/DualArmCoverCloth-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmCoverCloth-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.75", + "peak_step": "15000", + "final": "0.4", + "start": "2026-08-22 05:48:59", + "eval_mtime": "2026-08-25 10:46:28.520947", + "source_mtime": "2026-08-16 09:43:32.395415", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_dex/level_1/cloth_cover.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmCoverCloth-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmCoverCloth-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmCoverCloth-v1/lora_params/lora_step_0015000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmFlipTable-v1/eval_result.json b/previous_single_task/DualArmFlipTable-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..70a1d91a0984b1fae1e89d39218ca19b253e3190 --- /dev/null +++ b/previous_single_task/DualArmFlipTable-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmFlipTable-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_dex", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": 122.98795611858368, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmFlipTable-v1/evaluation_history.csv b/previous_single_task/DualArmFlipTable-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..48e7eac033b1e3c36fd0b130f7b706273c97a1eb --- /dev/null +++ b/previous_single_task/DualArmFlipTable-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,122.98795611858368,True +10000,50,0,0.0,-177.15296366930008,True +15000,50,0,0.0,-55.19636809825897,True +20000,50,0,0.0,547.2522929406166,True +25000,50,0,0.0,978.8827704238892,True +30000,50,0,0.0,1204.2270613360406,True +35000,50,0,0.0,624.2033227038384,True +40000,50,0,0.0,20.777628951072693,True diff --git a/previous_single_task/DualArmFlipTable-v1/lora_step_0005000.npz b/previous_single_task/DualArmFlipTable-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..a1b3f44c4f95c17c033c0e14f9fbe931f3075637 --- /dev/null +++ b/previous_single_task/DualArmFlipTable-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84f91077bad9c28bf3fd78e16d44e41b3de8a988f36eea6b6aa3346e44f7790a +size 1867943218 diff --git a/previous_single_task/DualArmFlipTable-v1/metadata.json b/previous_single_task/DualArmFlipTable-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..1d8640b01455ab27ea65caa5c33095c2a0c899a8 --- /dev/null +++ b/previous_single_task/DualArmFlipTable-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmFlipTable-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "84f91077bad9c28bf3fd78e16d44e41b3de8a988f36eea6b6aa3346e44f7790a", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "f05b0b8d3c15888a225f4bac65a28b4cba8fbc077413665a67428ce86251b92d", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmFlipTable-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_dex/level_4/flip_table.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmFlipTable-v1/norm_stats.json b/previous_single_task/DualArmFlipTable-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..578630b4c26bc04a4a6d0fd7da89a576b9f9f4bb --- /dev/null +++ b/previous_single_task/DualArmFlipTable-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0007426734082400799, + 0.0019271362107247114, + -0.004882488865405321, + 0.02135842852294445, + -0.01388007216155529, + 0.005700480192899704, + 0.0005747106624767184, + 0.018361659720540047, + 0.003528402652591467, + 0.0026970412582159042, + 0.00011019306111847982, + 0.0021281957160681486, + 0.00232951738871634, + 0.0020586117170751095, + 0.0021276555489748716, + 0.0031993589363992214, + 0.006803102791309357, + 0.004430961329489946, + 0.002992746653035283, + 0.001955868676304817, + 0.03019571490585804, + 0.007810966111719608, + 0.0005742975044995546, + -0.003479418810456991, + 0.004489129409193993, + -0.00449258740991354, + -0.000244803522946313, + -0.0005047322483733296, + 0.00038102324469946325, + 0.0001390924007864669, + 2.9386461619651527e-7, + 1.3414385158228015e-9, + -3.961932293350401e-7, + 0.00010006464435718954, + 3.2782881476123293e-7, + 1.613179101411788e-8, + -3.9956023556442233e-7, + 1.6740162891437649e-6, + -1.6116831602630555e-7, + -2.3883384869805013e-7, + -1.9461331248749048e-6, + 1.71792478909083e-7, + 3.187338393217942e-7, + 1.6119980905671127e-9 + ], + "std": [ + 0.036492493003606796, + 0.028516283258795738, + 0.035951826721429825, + 0.08862722665071487, + 0.07936225831508636, + 0.057571522891521454, + 0.0034002952743321657, + 0.062356408685445786, + 0.05329957604408264, + 0.05320403724908829, + 0.0016099086496978998, + 0.05334926396608353, + 0.05349767208099365, + 0.05350911244750023, + 0.0065283263102173805, + 0.0529230497777462, + 0.05411972850561142, + 0.05400650203227997, + 0.04093630984425545, + 0.00556001765653491, + 0.0801895409822464, + 0.05456719920039177, + 0.034291915595531464, + 0.025870798155665398, + 0.02653217874467373, + 0.0426500141620636, + 0.005235102027654648, + 0.012377274222671986, + 0.0022342156153172255, + 0.0008067899034358561, + 1.468961386308365e-6, + 1.1847170355849812e-8, + 3.3110427466453984e-6, + 0.0005726650124415755, + 1.685447614363511e-6, + 3.963235428727785e-7, + 4.565012659440981e-6, + 0.000015484800314879976, + 5.821274953632383e-6, + 2.673198878255789e-6, + 8.68780625751242e-6, + 3.7366512515291106e-6, + 6.816560016886797e-6, + 4.9021725345710365e-8 + ], + "q01": [ + -0.05999999865889549, + -0.05999999865889549, + -0.05999999865889549, + -0.012087200240790838, + -0.4000000059604645, + -0.07888000117540356, + -0.009981608675420283, + -0.15417153272628784, + -0.1989789619445801, + -0.20072442110180855, + -0.0008541274247691037, + -0.19841968781948088, + -0.19727040209770202, + -0.1972867751121521, + -0.00007455606684088692, + -0.1896670781850815, + -0.18928160519599918, + -0.19911767507791517, + -0.157510321110487, + -0.00008427028395235556, + -0.1176076498031616, + -0.1871109616279602, + -0.05999999865889549, + -0.05999999865889549, + -0.05999999865889549, + -0.4000000059604645, + -0.0015081045627594047, + -0.00008563443422315853, + 0.0, + 0.0, + -1.8644451242835198e-6, + -8.987250600966945e-9, + -0.0000138779375993181, + 0.0, + -1.8379588014795443e-6, + -1.0837117116579977e-8, + -0.000010893394648155673, + -0.000021105054573854437, + -0.000010769078310841012, + -5.604936335316786e-6, + -0.000039270104281604295, + -0.000018677513000147883, + -0.000034078314025100555, + -2.143525311680605e-7 + ], + "q99": [ + 0.05997599865943193, + 0.05997599865943193, + 0.05997599865943193, + 0.3999032945767045, + 0.019200000286102303, + 0.39984000595808034, + 0.00988946584016085, + 0.2297261057853699, + 0.2280834094703198, + 0.22801152437329297, + 0.0010817332779988645, + 0.2280854657292366, + 0.22805193715095518, + 0.22804641529917724, + 0.028380549090355638, + 0.22806448997259143, + 0.22812971086502065, + 0.22811115822196004, + 0.1753926777243614, + 0.01569595487769693, + 0.2280887239515782, + 0.22802289588451385, + 0.05997599865943193, + 0.05997599865943193, + 0.05997599865943193, + 0.0004000940471887149, + -6.695628166214496e-6, + 0.0008552481442690163, + 0.01321371151804924, + 0.004673562333919108, + 8.21313122060019e-6, + 6.766966376403615e-8, + 2.229006757261236e-6, + 0.0033424377545714376, + 9.173500243923628e-6, + 8.538533309661032e-8, + 2.188819542061547e-6, + 0.00005907930601388217, + 7.602139318623803e-6, + 2.98145826178528e-8, + 0.000019864914380013937, + 6.447361134087262e-6, + 0.000012014413380529728, + 1.2586919762611617e-7 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmFlipTable-v1/source_audit.json b/previous_single_task/DualArmFlipTable-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..dc083ad1560f70a1770c77a04354653f606b31db --- /dev/null +++ b/previous_single_task/DualArmFlipTable-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmFlipTable-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-09-04 06:54:15", + "eval_mtime": "2026-09-08 00:46:00.251516", + "source_mtime": "2026-08-28 06:58:09.348121", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_dex/level_4/flip_table.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmFlipTable-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmFlipTable-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmFlipTable-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmFoldHandKerchief-v1/eval_result.json b/previous_single_task/DualArmFoldHandKerchief-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..a955a9d4082a21cd474913330fee1a99db927c85 --- /dev/null +++ b/previous_single_task/DualArmFoldHandKerchief-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmFoldHandKerchief-v1", + "level": "level_1", + "suite": "deformable/dual_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 37, + "success_rate": 0.74, + "return_mean": -13.995554169416428, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmFoldHandKerchief-v1/evaluation_history.csv b/previous_single_task/DualArmFoldHandKerchief-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..7e72e7c30a5f14de5e9ed9e4612fd4c6ff67a9df --- /dev/null +++ b/previous_single_task/DualArmFoldHandKerchief-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,37,0.74,-13.995554169416428,True +10000,50,33,0.66,-23.69535655915737,True +15000,50,29,0.58,-23.868244435489178,True +20000,50,32,0.64,-25.854803668856622,True +25000,50,30,0.6,-21.692682106494903,True +30000,50,32,0.64,-17.018760781288147,True +35000,50,31,0.62,-25.837230781018732,True +40000,50,27,0.54,-19.315329426825045,True diff --git a/previous_single_task/DualArmFoldHandKerchief-v1/lora_step_0005000.npz b/previous_single_task/DualArmFoldHandKerchief-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..f48689c875773036dac64c760ee5b2d3161f7dcc --- /dev/null +++ b/previous_single_task/DualArmFoldHandKerchief-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:aae641e518457c83346e4ba2c617c3175eb9a18b6d8e3f2c526b20de921868aa +size 1867844866 diff --git a/previous_single_task/DualArmFoldHandKerchief-v1/metadata.json b/previous_single_task/DualArmFoldHandKerchief-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..b52991e3a5b886b0b44655d1af7fd0776078af6c --- /dev/null +++ b/previous_single_task/DualArmFoldHandKerchief-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmFoldHandKerchief-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "aae641e518457c83346e4ba2c617c3175eb9a18b6d8e3f2c526b20de921868aa", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "67faa683c6e21728a5df60d01867566ed1e6bb9cc1e03131b9e5bfdd4f2958c3", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmFoldHandKerchief-v1", + "source_path": "maniladder/envs/deformable/dual_arm_gripper/level_1/fold_handkerchief.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmFoldHandKerchief-v1/norm_stats.json b/previous_single_task/DualArmFoldHandKerchief-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..5cd69f2086a090bbd083ee56386db8b46fc0bf7c --- /dev/null +++ b/previous_single_task/DualArmFoldHandKerchief-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.01507443469017744, + -0.008877000771462917, + -0.04553778097033501, + 0.0, + 0.0, + 0.0, + -0.6149999499320984, + -0.00016005337238311768, + -0.021851276978850365, + -0.03744484856724739, + 0.0, + 0.0, + 0.0, + -0.022916661575436592, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.05923772230744362, + 0.048914410173892975, + 0.07224962115287781, + 0.0, + 0.0, + 0.0, + 0.7885271310806274, + 0.04140597581863403, + 0.05199756845831871, + 0.07513780146837234, + 0.0, + 0.0, + 0.0, + 0.9997373819351196, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.17499999701976776, + -0.17499999701976776, + -0.11500000208616257, + 0.0, + 0.0, + 0.0, + -1.0, + -0.17499999701976776, + -0.17499999701976776, + -0.11500999804139138, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.11494200208634137, + 0.056478002266585836, + 0.17494199701994656, + 0.0, + 0.0, + 0.0, + 0.9996, + 0.11486999804377557, + 0.11491246969401842, + 0.17492999702095985, + 0.0, + 0.0, + 0.0, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmFoldHandKerchief-v1/source_audit.json b/previous_single_task/DualArmFoldHandKerchief-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..1f8dff090e0a30cbf848736b7db2f0d505ae0ddf --- /dev/null +++ b/previous_single_task/DualArmFoldHandKerchief-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmFoldHandKerchief-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.95", + "peak_step": "5000", + "final": "0.7", + "start": "2026-08-21 15:10:28", + "eval_mtime": "2026-08-24 18:31:50.376622", + "source_mtime": "2026-08-10 07:33:18.457659", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_gripper/level_1/fold_handkerchief.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmFoldHandKerchief-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmFoldHandKerchief-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmFoldHandKerchief-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmJumpRope-v1/eval_result.json b/previous_single_task/DualArmJumpRope-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..f936743dfba54194fd932f1bd5ac2355b3270fb3 --- /dev/null +++ b/previous_single_task/DualArmJumpRope-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmJumpRope-v1", + "level": "level_4", + "suite": "deformable/dual_arm_dex", + "step": 20000, + "episodes": 50, + "successes": 1, + "success_rate": 0.02, + "return_mean": null, + "return_mean_finite": false, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmJumpRope-v1/evaluation_history.csv b/previous_single_task/DualArmJumpRope-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..9897064f34431e76a21256ae9348f320e7aaa71c --- /dev/null +++ b/previous_single_task/DualArmJumpRope-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,22.843300541203806,True +10000,50,0,0.0,,False +15000,50,0,0.0,,False +20000,50,1,0.02,,False +25000,50,0,0.0,241.48174225551077,True +30000,50,0,0.0,274.7488127852185,True +35000,50,1,0.02,309.5667886263051,True +40000,50,0,0.0,,False diff --git a/previous_single_task/DualArmJumpRope-v1/lora_step_0020000.npz b/previous_single_task/DualArmJumpRope-v1/lora_step_0020000.npz new file mode 100644 index 0000000000000000000000000000000000000000..3f26d4569ab4965343c39325278cfa5610018d26 --- /dev/null +++ b/previous_single_task/DualArmJumpRope-v1/lora_step_0020000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:15b86fb5ec34e19d2161c8f535d0ba3c2d0917f4b1434d574d1bb56ee7367487 +size 1867943218 diff --git a/previous_single_task/DualArmJumpRope-v1/metadata.json b/previous_single_task/DualArmJumpRope-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..3beb25ce1b7d6a85f3cdb11cc24b4d68e107e7a2 --- /dev/null +++ b/previous_single_task/DualArmJumpRope-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmJumpRope-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0020000.npz", + "checkpoint_sha256": "15b86fb5ec34e19d2161c8f535d0ba3c2d0917f4b1434d574d1bb56ee7367487", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 20000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "94b669285c03a24e8f37fb2d17964534a5005796c94f20a4751c365e1ffd3485", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmJumpRope-v1", + "source_path": "maniladder/envs/deformable/dual_arm_dex/level_4/jump_rope.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmJumpRope-v1/norm_stats.json b/previous_single_task/DualArmJumpRope-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..1da0d7e420d78ddec1d9af6284f94ed817f0eb7e --- /dev/null +++ b/previous_single_task/DualArmJumpRope-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.013891845010221004, + -0.005509881302714348, + 0.0007480590138584375, + 0.0, + 0.0, + 0.0, + 0.0004213227948639542, + 0.060306694358587265, + 0.02909696474671364, + 0.02337009645998478, + -0.0019281372660771012, + 0.044638488441705704, + 0.030073780566453934, + 0.030780363827943802, + 0.00029121257830411196, + 0.04038011282682419, + 0.02625971846282482, + 0.028399821370840073, + 0.02034723572432995, + -0.0030011325143277645, + 0.004695193842053413, + 0.03206956759095192, + -0.01687431149184704, + 0.010994810611009598, + 0.014131544157862663, + 0.0, + 0.0, + 0.0, + 0.001716078957542777, + 0.06603804230690002, + 0.03310388699173927, + 0.023476246744394302, + 0.002415557624772191, + 0.04284847527742386, + 0.026185709983110428, + 0.027835754677653313, + -0.0011805621907114983, + 0.03934244439005852, + 0.02654891088604927, + 0.029020817950367928, + -0.035112183541059494, + -0.0007496221805922687, + 0.01022364106029272, + 0.0343678779900074 + ], + "std": [ + 0.30494508147239685, + 0.07606858760118484, + 0.25489920377731323, + 0.0, + 0.0, + 0.0, + 0.0007292990921996534, + 0.14293478429317474, + 0.09175264835357666, + 0.1160225197672844, + 0.0016550853615626693, + 0.1477719247341156, + 0.09305509924888611, + 0.11503580957651138, + 0.0009371102205477655, + 0.14508768916130066, + 0.09229772537946701, + 0.11500002443790436, + 0.08607424050569534, + 0.007661766372621059, + 0.004273850470781326, + 0.13125446438789368, + 0.3065570294857025, + 0.06637248396873474, + 0.21719779074192047, + 0.0, + 0.0, + 0.0, + 0.0011479073436930776, + 0.141217440366745, + 0.09078820794820786, + 0.11595730483531952, + 0.0013141495874151587, + 0.1495560109615326, + 0.09355249255895615, + 0.11526011675596237, + 0.0006114624557085335, + 0.14673273265361786, + 0.09250997006893158, + 0.11494491249322891, + 0.0923219844698906, + 0.013791720382869244, + 0.005745499394834042, + 0.13127797842025757 + ], + "q01": [ + -1.25, + -0.12482455878257748, + -1.157, + 0.0, + 0.0, + 0.0, + -0.003084828959777951, + 0.0, + 0.0, + -0.0024270296562463045, + -0.003948463243618607, + 0.0, + 0.0, + 0.0, + -0.0023824935101903976, + -0.0002871990145649761, + 0.0, + 0.0, + -0.007452654037345201, + -0.00910922935884446, + -0.00004357305272642407, + 0.0, + -1.25, + -0.16542257933616633, + -0.4025949233293533, + 0.0, + 0.0, + 0.0, + -2.6316811272408813e-6, + 0.0, + 0.0, + -0.004092502407729626, + -1.0938762919977026e-6, + 0.0, + 0.0, + 0.0, + -0.0028168471799814142, + 0.0, + 0.0, + 0.0, + -0.497047283411026, + -0.06872799288202078, + -0.000047020820900797845, + 0.0 + ], + "q99": [ + 0.5255252298355104, + 0.46089046459198, + 0.42399999999999993, + 0.0, + 0.0, + 0.0, + 0.0014495970946736634, + 0.9998, + 0.6188286721348762, + 0.722528198191151, + 0.002117496396042407, + 0.9998, + 0.6226287811756134, + 0.8048000000000001, + 0.0026130862100049853, + 0.9997999425601971, + 0.6188289439201354, + 0.8048000000000001, + 0.49698482463378457, + 0.03379347009118647, + 0.024084568159537596, + 0.9944000000000001, + 0.5211171898841858, + 0.16480248024463662, + 0.4267901689767837, + 0.0, + 0.0, + 0.0, + 0.0035142822094785518, + 0.9998, + 0.6195859881877899, + 0.7224688323345035, + 0.0051485651440278165, + 0.9998, + 0.626909701704979, + 0.8048000000000001, + 0.00027408831355860445, + 0.9998, + 0.6211425185799598, + 0.8048000000000001, + -0.00016998881101615293, + 0.006253963422589007, + 0.0318287375997752, + 0.9944000000000001 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmJumpRope-v1/source_audit.json b/previous_single_task/DualArmJumpRope-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..4f4aff955e1b4688c24a5f18b7bc1cf80a4a026d --- /dev/null +++ b/previous_single_task/DualArmJumpRope-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmJumpRope-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.05", + "peak_step": "40000", + "final": "0.05", + "start": "2026-09-04 06:14:20", + "eval_mtime": "2026-09-07 19:21:22.813556", + "source_mtime": "2026-07-24 05:42:56.217141", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_dex/level_4/jump_rope.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmJumpRope-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmJumpRope-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmJumpRope-v1/lora_params/lora_step_0040000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmKeyboardType-v1/eval_result.json b/previous_single_task/DualArmKeyboardType-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..57a90caf89032465a6f067a85f290396c6207f8c --- /dev/null +++ b/previous_single_task/DualArmKeyboardType-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmKeyboardType-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_dex", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": -248.3057014244236, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 7, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmKeyboardType-v1/evaluation_history.csv b/previous_single_task/DualArmKeyboardType-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..4df15cb744d836f0f7e5cf8bef82f3bf5474a2e9 --- /dev/null +++ b/previous_single_task/DualArmKeyboardType-v1/evaluation_history.csv @@ -0,0 +1,8 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,-248.3057014244236,True +10000,50,0,0.0,-237.66442889969795,True +20000,50,0,0.0,-246.98060533750336,True +25000,50,0,0.0,-230.77981589608186,True +30000,50,0,0.0,-244.4625939191133,True +35000,50,0,0.0,-246.3290418352792,True +40000,50,0,0.0,-243.75864611959085,True diff --git a/previous_single_task/DualArmKeyboardType-v1/lora_step_0005000.npz b/previous_single_task/DualArmKeyboardType-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..3e504907bcec13624262a07cfa46c6460802ac8b --- /dev/null +++ b/previous_single_task/DualArmKeyboardType-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:521d586261c29c45cf239aebd59190737d7be7d05cc1afd943f361e72fa76bee +size 1867943218 diff --git a/previous_single_task/DualArmKeyboardType-v1/metadata.json b/previous_single_task/DualArmKeyboardType-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..284570400a0b9153d230be8771d956e2295180f4 --- /dev/null +++ b/previous_single_task/DualArmKeyboardType-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmKeyboardType-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "521d586261c29c45cf239aebd59190737d7be7d05cc1afd943f361e72fa76bee", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "dd2b09fa6c67705171d217c05c94573faeda7662ed544fc0d0a102e4e6e3573f", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmKeyboardType-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_dex/level_4/keyboard_type.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmKeyboardType-v1/norm_stats.json b/previous_single_task/DualArmKeyboardType-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..dcda0d8e46d00fa7ff83fe60085ce45020956899 --- /dev/null +++ b/previous_single_task/DualArmKeyboardType-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0015773044433444738, + -0.010229155421257019, + -0.00017001946980599314, + 0.0, + 0.0, + 0.0, + 0.0, + -7.096505674764053e-12, + -1.4193011349528106e-11, + -2.8386022699056213e-11, + 0.0, + 1.227613930648852e-11, + 2.455227861297704e-11, + 4.910455722595408e-11, + 0.0, + -3.1475449416978885e-12, + -6.295089883395777e-12, + -1.2590179766791554e-11, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0018256823532283306, + 0.0010209102183580399, + -0.0013404387282207608, + 0.0, + 0.0, + 0.0, + 0.0, + -1.2675827401609219e-11, + -2.5351654803218437e-11, + -5.0703309606436875e-11, + 0.0, + 1.4551915228366852e-11, + 2.9103830456733704e-11, + 5.820766091346741e-11, + 0.0, + 7.275957614183426e-12, + 1.4551915228366852e-11, + 2.9103830456733704e-11, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.01555612962692976, + 0.02666105516254902, + 0.026018748059868813, + 0.0, + 0.0, + 0.0, + 0.0, + 0.025799976661801338, + 0.051599953323602676, + 0.10319990664720535, + 0.0, + 0.022923875600099564, + 0.04584775120019913, + 0.09169550240039825, + 0.0, + 0.03696364909410477, + 0.07392729818820953, + 0.14785459637641907, + 0.0, + 0.0, + 0.0, + 0.0, + 0.026178857311606407, + 0.026960931718349457, + 0.030237974599003792, + 0.0, + 0.0, + 0.0, + 0.0, + 0.041432447731494904, + 0.08286489546298981, + 0.16572979092597961, + 0.0, + 0.03501681610941887, + 0.07003363221883774, + 0.14006726443767548, + 0.0, + 0.03348247706890106, + 0.06696495413780212, + 0.13392990827560425, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.05000000074505806, + -0.05000000074505806, + -0.05000000074505806, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + -0.25, + -0.5, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + -0.05000000074505806, + -0.05000000074505806, + -0.05000000074505806, + 0.0, + 0.0, + 0.0, + 0.0, + -0.25, + -0.5, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.04998000074476004, + 0.04998000074476004, + 0.04998000074476004, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.2499, + 0.4998, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.04998000074476004, + 0.04998000074476004, + 0.04998000074476004, + 0.0, + 0.0, + 0.0, + 0.0, + 0.2499, + 0.4998, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmKeyboardType-v1/source_audit.json b/previous_single_task/DualArmKeyboardType-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..a17003b248d606164caf239e82b010d348c16007 --- /dev/null +++ b/previous_single_task/DualArmKeyboardType-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmKeyboardType-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 17:55:07", + "eval_mtime": "2026-08-27 03:05:42.733629", + "source_mtime": "2026-07-02 12:37:44.741583", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_dex/level_4/keyboard_type.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmKeyboardType-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmKeyboardType-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmKeyboardType-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmMahjongSetup-v1/eval_result.json b/previous_single_task/DualArmMahjongSetup-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..561369739419805dc4b8c9be59637bda578788a1 --- /dev/null +++ b/previous_single_task/DualArmMahjongSetup-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmMahjongSetup-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 1, + "success_rate": 0.02, + "return_mean": null, + "return_mean_finite": false, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmMahjongSetup-v1/evaluation_history.csv b/previous_single_task/DualArmMahjongSetup-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..83eb7ee404b5be7b16d574fd5e65f567484659bf --- /dev/null +++ b/previous_single_task/DualArmMahjongSetup-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,1,0.02,,False +10000,50,0,0.0,,False +15000,50,0,0.0,,False +20000,50,0,0.0,,False +25000,50,0,0.0,,False +30000,50,0,0.0,,False +35000,50,0,0.0,,False +40000,50,0,0.0,,False diff --git a/previous_single_task/DualArmMahjongSetup-v1/lora_step_0005000.npz b/previous_single_task/DualArmMahjongSetup-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..84caa30e28ffef2a41cb81a522d6b6c961ba9096 --- /dev/null +++ b/previous_single_task/DualArmMahjongSetup-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f565771d810c4e9c7d1499feee6ffbe540235cabab920b80634365b74c11a51a +size 1867844866 diff --git a/previous_single_task/DualArmMahjongSetup-v1/metadata.json b/previous_single_task/DualArmMahjongSetup-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..b0d9b7d6e761669f72afe4e0a14be134397d5c77 --- /dev/null +++ b/previous_single_task/DualArmMahjongSetup-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmMahjongSetup-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "f565771d810c4e9c7d1499feee6ffbe540235cabab920b80634365b74c11a51a", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "2e10311ca007529deb8c27caab85723c2c69a2139ddb34ad4de61eb06db6afb6", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmMahjongSetup-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_gripper/level_4/mahjong_setup.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmMahjongSetup-v1/norm_stats.json b/previous_single_task/DualArmMahjongSetup-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..50a19252a553f770af57009e7248ec2d10080c56 --- /dev/null +++ b/previous_single_task/DualArmMahjongSetup-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0035207634791731834, + -0.01898203231394291, + -0.010802083648741245, + 0.005854140501469374, + -0.01160410139709711, + -0.002055011224001646, + 0.3861602544784546, + 0.003475473029538989, + -0.00204411125741899, + -0.005668812431395054, + -0.004295963328331709, + -0.011232728138566017, + 0.0021876327227801085, + -0.020004838705062866, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.01593903824687004, + 0.04697767645120621, + 0.07459336519241333, + 0.024683335795998573, + 0.021006735041737556, + 0.004511654376983643, + 0.7359713315963745, + 0.06542954593896866, + 0.11066830903291702, + 0.16598959267139435, + 0.0351712666451931, + 0.031552717089653015, + 0.10051490366458893, + 0.8308870792388916, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.030410898113250734, + -0.30000001192092896, + -0.30000001192092896, + -0.0005893052220344586, + -0.05000000074505806, + -0.01813647662848234, + -1.0, + -0.30000001192092896, + -0.30000001192092896, + -0.30000001192092896, + -0.20000000298023224, + -0.07664000114202499, + -0.20000000298023224, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.03831813608705997, + 0.019200000762939484, + 0.074900021597743, + 0.1999332511305809, + 0.0038069832898676437, + 0.0030534490719437563, + 0.9996, + 0.29988001191616065, + 0.29988001191616065, + 0.29988001191616065, + 0.09960000148415565, + 0.07904000117778781, + 0.19992000297904017, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmMahjongSetup-v1/source_audit.json b/previous_single_task/DualArmMahjongSetup-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..1832738fcc71a27ae562c9566335f79d6be8df48 --- /dev/null +++ b/previous_single_task/DualArmMahjongSetup-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmMahjongSetup-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 06:21:00", + "eval_mtime": "2026-08-27 05:07:51.486450", + "source_mtime": "2026-07-06 15:41:30.671964", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_gripper/level_4/mahjong_setup.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmMahjongSetup-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmMahjongSetup-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmMahjongSetup-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmOpenBottle-v1/eval_result.json b/previous_single_task/DualArmOpenBottle-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..56f0d4ae164dac026a397172e62e85be16f2cd4b --- /dev/null +++ b/previous_single_task/DualArmOpenBottle-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmOpenBottle-v1", + "level": "level_2", + "suite": "rigid_body/dual_arm_dex", + "step": 30000, + "episodes": 50, + "successes": 3, + "success_rate": 0.06, + "return_mean": 2172.774367170334, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmOpenBottle-v1/evaluation_history.csv b/previous_single_task/DualArmOpenBottle-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..2ece0a6b8d7a056fa31764c906c4822a850208d4 --- /dev/null +++ b/previous_single_task/DualArmOpenBottle-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,2160.166076388359,True +10000,50,0,0.0,2153.5597602939606,True +15000,50,0,0.0,2186.192955737114,True +20000,50,1,0.02,2164.021308274269,True +25000,50,1,0.02,2198.19865986824,True +30000,50,3,0.06,2172.774367170334,True +35000,50,2,0.04,2172.3835971689223,True +40000,50,1,0.02,2156.116339635849,True diff --git a/previous_single_task/DualArmOpenBottle-v1/lora_step_0030000.npz b/previous_single_task/DualArmOpenBottle-v1/lora_step_0030000.npz new file mode 100644 index 0000000000000000000000000000000000000000..b1be8db35b3c513047cfb1896831468e66a1d4e2 --- /dev/null +++ b/previous_single_task/DualArmOpenBottle-v1/lora_step_0030000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:badce300065df7e4c45bb4cfb07dfcf74dcd6a33c524f6de76ef3a15aff2490f +size 1867943218 diff --git a/previous_single_task/DualArmOpenBottle-v1/metadata.json b/previous_single_task/DualArmOpenBottle-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..83f418981d0a28e619be156722715dc070392927 --- /dev/null +++ b/previous_single_task/DualArmOpenBottle-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmOpenBottle-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0030000.npz", + "checkpoint_sha256": "badce300065df7e4c45bb4cfb07dfcf74dcd6a33c524f6de76ef3a15aff2490f", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 30000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "038f653494878bc3d401547bcc9a9cf4f4709c7667eafec5fdd06bc77895b61f", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmOpenBottle-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_dex/level_2/open_bottle.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmOpenBottle-v1/norm_stats.json b/previous_single_task/DualArmOpenBottle-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..43f696e01d6bf0112ef527bb1f3e1f1792a30748 --- /dev/null +++ b/previous_single_task/DualArmOpenBottle-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0030260083731263876, + 0.011173166334629059, + -0.0036629943642765284, + -0.00031854090048000216, + -0.00042784938705153763, + -0.001185431843623519, + -0.0024862014688551426, + 0.03064289689064026, + 0.027146417647600174, + 0.020708447322249413, + -0.0014122138964012265, + 0.023419788107275963, + 0.021622972562909126, + 0.017870936542749405, + 0.0017429656581953168, + 0.024516725912690163, + 0.02234458178281784, + 0.018197963014245033, + 0.03447521850466728, + -9.174062256533944e-7, + -1.65206290603237e-6, + -4.501061467987455e-10, + 0.004066111985594034, + -0.010334658436477184, + 0.016413340345025063, + 0.0006609841366298497, + -0.00029635048122145236, + 0.002523430623114109, + -1.651539719205175e-6, + 0.0002478914684616029, + 3.597776583319501e-7, + -9.073149453975304e-10, + -1.7897726820592652e-6, + 3.178890665367362e-6, + 4.2225303786835866e-7, + -1.0434333397313367e-9, + -1.689935743343085e-6, + 3.5409489100857172e-6, + 4.67635516088194e-7, + -1.0485495804957168e-9, + 0.06302130222320557, + -0.0032394297886639833, + 0.039036788046360016, + 0.030170373618602753 + ], + "std": [ + 0.020752672106027603, + 0.026943575590848923, + 0.019242262467741966, + 0.0015494675608351827, + 0.0012767129810526967, + 0.0032808990217745304, + 0.0018719973741099238, + 0.03537898510694504, + 0.035808611661195755, + 0.037032052874565125, + 0.0012087657814845443, + 0.036421455442905426, + 0.03678526729345322, + 0.03769497573375702, + 0.0019598272629082203, + 0.036286596208810806, + 0.03668608143925667, + 0.03763090446591377, + 0.08828607946634293, + 2.134521082552965e-6, + 3.841631951218005e-6, + 4.879240123045747e-9, + 0.042572181671857834, + 0.028351837769150734, + 0.03212401643395424, + 0.04033517837524414, + 0.0014014104381203651, + 0.008475938811898232, + 4.864918537350604e-6, + 0.0006467134226113558, + 8.028056868170097e-7, + 2.2563901858063673e-9, + 3.4279100873391144e-6, + 6.161598321341444e-6, + 7.5960696221955e-7, + 2.3113166935928575e-9, + 3.521969802022795e-6, + 6.401816335710464e-6, + 7.872240530559793e-7, + 2.3091957235266136e-9, + 0.10396398603916168, + 0.014436093159019947, + 0.07085937261581421, + 0.06255413591861725 + ], + "q01": [ + -0.024287003138661382, + -0.00582399986982346, + -0.07999999821186066, + -0.0037034088768064977, + -0.003806878996640444, + -0.012106861859932542, + -0.006364170050516259, + -0.000029854903914383613, + -3.618056325649377e-6, + -1.0377585901721886e-8, + -0.003587132411357015, + -0.000029749315217486583, + -3.603935283535975e-6, + -1.0374453296435604e-8, + -0.00036870941906236114, + -0.00002964156556117814, + -3.588673280319199e-6, + -1.037114572000064e-8, + -0.0020145856886170806, + -8.393864713923448e-6, + -0.000015100344743791539, + -2.0588399976873007e-8, + -0.10000000149011612, + -0.10000000149011612, + -0.009793867102265359, + -0.10452676876485345, + -0.006354101292788982, + -0.003964626462012531, + -0.000012082354266567563, + -0.000137773498945171, + -6.218778825314081e-7, + -8.157977833977981e-9, + -0.000012065031023666961, + -1.5269672483555045e-6, + -2.48761833199751e-7, + -7.341386006487482e-9, + -0.000012110039778417559, + -2.417580448673107e-6, + -3.572280309072085e-7, + -7.348869752377141e-9, + -0.00046754557462409135, + -0.07312034339271486, + -0.00001455734673072584, + -2.7346134245931353e-8 + ], + "q99": [ + 0.07997825816571713, + 0.07996799821257591, + 0.010090616860985765, + 0.0023668251477181905, + 0.002172964123263954, + 0.005582260824181139, + 2.0317996386395865e-7, + 0.2044025571174847, + 0.20442499272399509, + 0.20442800226020102, + 0.00022735570028889918, + 0.2044027984064931, + 0.20442501992949746, + 0.20442800226071978, + 0.008033262569445651, + 0.20440306492041418, + 0.204425047323943, + 0.2044280022612675, + 0.4591148227545898, + -1.6623328633616962e-9, + -2.736922897383009e-9, + 1.3689184128651278e-8, + 0.09996000148952008, + 0.01700000025331974, + 0.07998010149896144, + 0.015525775477290171, + 0.0022769956104457395, + 0.03936893504932522, + 4.462925194638955e-6, + 0.002906881181983044, + 2.552589511742553e-6, + 3.1574207413065393e-9, + 1.5077605849000957e-6, + 0.000021476102936139795, + 2.66530514400074e-6, + 3.0941987088084494e-9, + 2.6224380508210733e-6, + 0.000022505880909193364, + 2.791098075931586e-6, + 3.0387376810381276e-9, + 0.2945361961950781, + 0.012106769988313323, + 0.3135137393521472, + 0.3079550258840101 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmOpenBottle-v1/source_audit.json b/previous_single_task/DualArmOpenBottle-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..0e166443206facd9d49054ce66f0bbe2e8f76e06 --- /dev/null +++ b/previous_single_task/DualArmOpenBottle-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmOpenBottle-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.1", + "peak_step": "30000", + "final": "0.05", + "start": "2026-08-22 06:01:06", + "eval_mtime": "2026-08-27 00:37:33.419041", + "source_mtime": "2026-05-23 06:54:07.598059", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_dex/level_2/open_bottle.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmOpenBottle-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmOpenBottle-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmOpenBottle-v1/lora_params/lora_step_0030000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmOpenFridge-v1/eval_result.json b/previous_single_task/DualArmOpenFridge-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..dc4c63aa2607557fa3c97f4c7f75eae1a9c7f826 --- /dev/null +++ b/previous_single_task/DualArmOpenFridge-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmOpenFridge-v1", + "level": "level_3", + "suite": "rigid_body/dual_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": 1425.2416665172577, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmOpenFridge-v1/evaluation_history.csv b/previous_single_task/DualArmOpenFridge-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..aa4e4eca802dd684464792c67f20e64a345f1122 --- /dev/null +++ b/previous_single_task/DualArmOpenFridge-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,1425.2416665172577,True +10000,50,0,0.0,1425.2351796388625,True +15000,50,0,0.0,1425.235145483017,True +20000,50,0,0.0,1425.2351254606247,True +25000,50,0,0.0,1425.2351526212692,True +30000,50,0,0.0,1425.235009675026,True +35000,50,0,0.0,1425.2351335525514,True +40000,50,0,0.0,1425.2403667545318,True diff --git a/previous_single_task/DualArmOpenFridge-v1/lora_step_0005000.npz b/previous_single_task/DualArmOpenFridge-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..c15159b342d1d6775ef83ddf6148a08d5a985d70 --- /dev/null +++ b/previous_single_task/DualArmOpenFridge-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:311033395a4c57babcae115e5a0ae2c179504482d1d85075956fd03d545c5e48 +size 1867844866 diff --git a/previous_single_task/DualArmOpenFridge-v1/metadata.json b/previous_single_task/DualArmOpenFridge-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..d4dbd8f79ac0c3b1187d0c7f8ec7ce2d44087777 --- /dev/null +++ b/previous_single_task/DualArmOpenFridge-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmOpenFridge-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "311033395a4c57babcae115e5a0ae2c179504482d1d85075956fd03d545c5e48", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "bbb6b5e39f0a771a54315536c2fd8d6aa04f7a9e5f1e79cce56d459f8ca4cd31", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmOpenFridge-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_gripper/level_3/open_fridge.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmOpenFridge-v1/norm_stats.json b/previous_single_task/DualArmOpenFridge-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..39d2e374c43a6a2ca06db50542ad3105ddbf685a --- /dev/null +++ b/previous_single_task/DualArmOpenFridge-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.4416833817958832, + -0.50259929895401, + -3.307809829711914, + -0.0055829561315476894, + 0.004735257476568222, + -0.0018423235742375255, + -0.0002783526433631778, + 0.00317446724511683, + 0.002320999978110194, + -0.12007831037044525, + -0.4739932417869568, + -0.7375828623771667, + -2.742861270904541, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.33034753799438477, + 0.5980759859085083, + 0.5772335529327393, + 0.0386582650244236, + 0.027031777426600456, + 0.032965824007987976, + 0.002946489257737994, + 0.012892046011984348, + 0.010310106910765171, + 0.992764413356781, + 0.2330041527748108, + 0.6563135981559753, + 0.9742379784584045, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.9244334932565689, + -1.758625662112236, + -4.71314493150711, + -0.11999999731779099, + -0.047663998934626584, + -0.11999999731779099, + -0.015026142785325644, + -0.004638719291985033, + -0.0008726208969950678, + -1.0, + -0.6790909767150879, + -1.2404043707847596, + -3.145308067369461, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.12356583552360534, + 0.4995461873292921, + -1.713554708957672, + 0.11995199731886386, + 0.11995199731886386, + 0.11995199731886386, + 0.0004791170340031417, + 0.05998335635252297, + 0.059987013885099444, + 0.9996, + -0.00012730304108121704, + 0.49965584135055563, + -0.0006620307445523999, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmOpenFridge-v1/source_audit.json b/previous_single_task/DualArmOpenFridge-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..658867823d6f6cb14f76ce57753fa748ce78162d --- /dev/null +++ b/previous_single_task/DualArmOpenFridge-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmOpenFridge-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 06:21:02", + "eval_mtime": "2026-08-25 06:23:24.764906", + "source_mtime": "2026-08-10 09:04:07.997708", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_gripper/level_3/open_fridge.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmOpenFridge-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmOpenFridge-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmOpenFridge-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmOpenLaptop-v1/eval_result.json b/previous_single_task/DualArmOpenLaptop-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..a374a32761ef1a1cfc564777b383e8ae06ea93a6 --- /dev/null +++ b/previous_single_task/DualArmOpenLaptop-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmOpenLaptop-v1", + "level": "level_2", + "suite": "rigid_body/dual_arm_dex", + "step": 25000, + "episodes": 50, + "successes": 21, + "success_rate": 0.42, + "return_mean": 1361.7338301420211, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmOpenLaptop-v1/evaluation_history.csv b/previous_single_task/DualArmOpenLaptop-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..bf0b5902c481df3894e77911183189fd204ed649 --- /dev/null +++ b/previous_single_task/DualArmOpenLaptop-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,3,0.06,1662.1150017833709,True +10000,50,7,0.14,1569.3558869957924,True +15000,50,15,0.3,1423.962103497982,True +20000,50,14,0.28,1446.9925638628006,True +25000,50,21,0.42,1361.7338301420211,True +30000,50,17,0.34,1400.763299677372,True +35000,50,14,0.28,1430.9343499541283,True +40000,50,18,0.36,1398.3386854696273,True diff --git a/previous_single_task/DualArmOpenLaptop-v1/lora_step_0025000.npz b/previous_single_task/DualArmOpenLaptop-v1/lora_step_0025000.npz new file mode 100644 index 0000000000000000000000000000000000000000..9e3ac91ebd86d5ddfbfe1e60fa1f6280d9daf324 --- /dev/null +++ b/previous_single_task/DualArmOpenLaptop-v1/lora_step_0025000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:96e016a4d0628be868f5ae1108397addc4216fb3123020171775346ad335cca3 +size 1867943218 diff --git a/previous_single_task/DualArmOpenLaptop-v1/metadata.json b/previous_single_task/DualArmOpenLaptop-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..452815b44daff77d0727960406487c6c21e70e9f --- /dev/null +++ b/previous_single_task/DualArmOpenLaptop-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmOpenLaptop-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0025000.npz", + "checkpoint_sha256": "96e016a4d0628be868f5ae1108397addc4216fb3123020171775346ad335cca3", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 25000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "11adf753b20cac6c6ea0f5b7740d16201d0deb56d17e3e0aa40299f0732236be", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmOpenLaptop-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_dex/level_2/open_laptop.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmOpenLaptop-v1/norm_stats.json b/previous_single_task/DualArmOpenLaptop-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..1da0808a740e4d7233f1012e1ef6e28658028d87 --- /dev/null +++ b/previous_single_task/DualArmOpenLaptop-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.009761474095284939, + 0.0030048973858356476, + -0.022122660651803017, + 0.0, + 0.0, + 0.0, + 0.00315500283613801, + 0.029833251610398293, + 0.022331584244966507, + 0.013039285317063332, + -0.00021284985996317118, + 0.011756954714655876, + 0.003886700375005603, + 0.00479178037494421, + 0.001190439099445939, + 0.016575701534748077, + 0.0024542079772800207, + 0.004806192126125097, + 0.046735797077417374, + 0.029040502384305, + 0.0744381994009018, + 0.051169540733098984, + 0.007095549255609512, + -0.011560690589249134, + -0.032358817756175995, + 0.0, + 0.0, + 0.0, + 0.00020570879860315472, + 0.005589714739471674, + 0.007380764000117779, + 0.007890021428465843, + 0.000021451296561281197, + 0.0049300240352749825, + 0.007869274355471134, + 0.007909062318503857, + -0.00048616150161251426, + 0.005815297365188599, + 0.00816378090530634, + 0.008004915900528431, + 0.18647412955760956, + -0.0002965758030768484, + 0.07134260982275009, + 0.05193725973367691 + ], + "std": [ + 0.05045509338378906, + 0.05751597136259079, + 0.05881240591406822, + 0.0, + 0.0, + 0.0, + 0.003913707099854946, + 0.03676508739590645, + 0.028989743441343307, + 0.025364534929394722, + 0.0012355971848592162, + 0.03750746324658394, + 0.025818800553679466, + 0.025225915014743805, + 0.006887446157634258, + 0.03949761390686035, + 0.026030661538243294, + 0.025222016498446465, + 0.08124406635761261, + 0.02779923938214779, + 0.06972640007734299, + 0.04944319650530815, + 0.028076516464352608, + 0.06265799701213837, + 0.060308150947093964, + 0.0, + 0.0, + 0.0, + 0.0029026444535702467, + 0.05419394373893738, + 0.03879496455192566, + 0.038581542670726776, + 0.0003926566569134593, + 0.05398574098944664, + 0.03858582675457001, + 0.03857704624533653, + 0.0026566488668322563, + 0.05401305854320526, + 0.03856729343533516, + 0.038566429167985916, + 0.18657435476779938, + 0.0018213932635262609, + 0.06957893073558807, + 0.054409999400377274 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -0.006691354110836982, + -0.00038916758960112936, + -0.00012052637115120626, + -0.00007228788882493974, + -0.006264054720895365, + -0.011311900329589841, + -0.02367546048462391, + -0.00003728416068479387, + -0.012593900150060653, + -0.014303311997652054, + -0.03194283764958382, + -3.138039605943277e-9, + -5.461963155539706e-6, + -0.000016750444046920165, + -1.9760905161092523e-6, + -2.555977118845476e-8, + -0.04128000061511993, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -0.004836076673865319, + -0.16905151122808457, + -0.02025882121324539, + -0.000024525988847017358, + -0.0001293191919103263, + -0.16905017906427383, + -0.0017318841077387332, + -0.00024777639051899313, + -0.014932851097732782, + -0.16778354756236075, + -0.002996089961379766, + -0.0017400672663003209, + -0.08648919350504874, + -0.005119291858747601, + -0.00006506888894364238, + -8.738879841985181e-8 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + 0.011898991145193577, + 0.20432875934708863, + 0.13205748835094275, + 0.1322013316531782, + 0.0015841787479817862, + 0.2040944017916918, + 0.13190418414622543, + 0.13216515290308745, + 0.03871319642141462, + 0.20393604145646094, + 0.1320052585706115, + 0.13222300158451886, + 0.5099312935494512, + 0.056763476776132304, + 0.1465086644330231, + 0.13595398599565936, + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + 0.015182535378634926, + 0.23651217517256734, + 0.21996186988502742, + 0.21993470465391876, + 0.0012787035666406155, + 0.23651264157891266, + 0.2199960554998368, + 0.21998632458373904, + 0.0021958319023251544, + 0.2364745692372322, + 0.2199536587111652, + 0.21995517836026848, + 0.6091305584222078, + 0.0033273817934095853, + 0.22000440476720687, + 0.21994696119378151 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmOpenLaptop-v1/source_audit.json b/previous_single_task/DualArmOpenLaptop-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..8a4e7625ea63ce6a23737bc680b8493955fa02cb --- /dev/null +++ b/previous_single_task/DualArmOpenLaptop-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmOpenLaptop-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.45", + "peak_step": "25000", + "final": "0.25", + "start": "2026-08-22 06:31:00", + "eval_mtime": "2026-08-25 11:20:14.985634", + "source_mtime": "2026-05-22 16:10:36.791582", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_dex/level_2/open_laptop.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmOpenLaptop-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmOpenLaptop-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmOpenLaptop-v1/lora_params/lora_step_0025000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmOpenPen-v1/eval_result.json b/previous_single_task/DualArmOpenPen-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..816e92ed49addb13c005f704adcb13ecc233281e --- /dev/null +++ b/previous_single_task/DualArmOpenPen-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmOpenPen-v1", + "level": "level_1", + "suite": "rigid_body/dual_arm_dex", + "step": 35000, + "episodes": 50, + "successes": 18, + "success_rate": 0.36, + "return_mean": 798.6847934004664, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmOpenPen-v1/evaluation_history.csv b/previous_single_task/DualArmOpenPen-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..0c36c91054eda7c1fb75e36d85bd4d820b008978 --- /dev/null +++ b/previous_single_task/DualArmOpenPen-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,17,0.34,848.7281628412009,True +10000,50,16,0.32,837.5850449207425,True +15000,50,10,0.2,805.6612854599953,True +20000,50,17,0.34,837.3289239484071,True +25000,50,11,0.22,,False +30000,50,11,0.22,826.5384598797559,True +35000,50,18,0.36,798.6847934004664,True +40000,50,14,0.28,,False diff --git a/previous_single_task/DualArmOpenPen-v1/lora_step_0035000.npz b/previous_single_task/DualArmOpenPen-v1/lora_step_0035000.npz new file mode 100644 index 0000000000000000000000000000000000000000..edb7a47ffb328f55b589932d2bf5bcdada3dc80c --- /dev/null +++ b/previous_single_task/DualArmOpenPen-v1/lora_step_0035000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8057718dae051ffe7dd5641fc5e50f135068359806d80cd29a273349c008dd78 +size 1867943218 diff --git a/previous_single_task/DualArmOpenPen-v1/metadata.json b/previous_single_task/DualArmOpenPen-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..833e7092728e7362f06220b16a5e76f4cc0ccf87 --- /dev/null +++ b/previous_single_task/DualArmOpenPen-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmOpenPen-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0035000.npz", + "checkpoint_sha256": "8057718dae051ffe7dd5641fc5e50f135068359806d80cd29a273349c008dd78", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 35000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "4750165c869400c1057312a4f8d35e834a6f159fda3b619c6170c8a175696f3d", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmOpenPen-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_dex/level_1/open_pen.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmOpenPen-v1/norm_stats.json b/previous_single_task/DualArmOpenPen-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..a3c88c52188b3da54ac0ed32af94a0be13b3c296 --- /dev/null +++ b/previous_single_task/DualArmOpenPen-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.02282295562326908, + 0.015205384232103825, + -0.008606472983956337, + 0.0, + 0.0, + 0.0, + 0.001909521990455687, + 0.059554323554039, + 0.06017771363258362, + 0.0544203445315361, + 0.001227720407769084, + 0.05552687123417854, + 0.0561472550034523, + 0.05165986716747284, + 0.001356413122266531, + 0.052750829607248306, + 0.05301051214337349, + 0.050130896270275116, + 0.06012676656246185, + 0.017094686627388, + 0.0744996890425682, + 0.06626155972480774, + 0.023735657334327698, + -0.009175165556371212, + -0.010348314419388771, + 0.0, + 0.0, + 0.0, + 0.0004127856809645891, + 0.04896827042102814, + 0.049023862928152084, + 0.04774896800518036, + 0.00010678276885300875, + 0.04736277833580971, + 0.04757707193493843, + 0.04693278670310974, + 0.000045628628868144006, + 0.04687045142054558, + 0.046902433037757874, + 0.046552080661058426, + 0.10246549546718597, + -0.0005973735824227333, + 0.07704651355743408, + 0.06617273390293121 + ], + "std": [ + 0.02934980019927025, + 0.03287196904420853, + 0.024170471355319023, + 0.0, + 0.0, + 0.0, + 0.01854347065091133, + 0.08572494983673096, + 0.08416856825351715, + 0.08332876861095428, + 0.01316402480006218, + 0.0830790176987648, + 0.08366376161575317, + 0.08340954035520554, + 0.011258621700108051, + 0.08403708785772324, + 0.08429055660963058, + 0.08414207398891449, + 0.0713319480419159, + 0.02651394158601761, + 0.0883750319480896, + 0.08486523479223251, + 0.025905901566147804, + 0.0270552858710289, + 0.020349860191345215, + 0.0, + 0.0, + 0.0, + 0.0016766928602010012, + 0.08395271748304367, + 0.08413100987672806, + 0.08409811556339264, + 0.00041115665226243436, + 0.08393847942352295, + 0.08392954617738724, + 0.08408845216035843, + 0.00016956769104581326, + 0.0840601995587349, + 0.0840761810541153, + 0.0841846838593483, + 0.10089706629514694, + 0.005048058927059174, + 0.0918046236038208, + 0.08643291145563126 + ], + "q01": [ + -0.05000000074505806, + -0.05000000074505806, + -0.05000000074505806, + 0.0, + 0.0, + 0.0, + -0.01860996693968773, + 0.0, + 0.0, + 0.0, + -0.0038401078367605807, + 0.0, + 0.0, + 0.0, + -0.0014698280016891658, + 0.0, + 0.0, + 0.0, + 0.0, + -0.00006728609129786484, + 0.0, + 0.0, + -0.041600000619888305, + -0.05000000074505806, + -0.05000000074505806, + 0.0, + 0.0, + 0.0, + -0.00005399122752714902, + 0.0, + 0.0, + 0.0, + -0.0001326213899301365, + 0.0, + 0.0, + 0.0, + -0.000011462177350040293, + 0.0, + 0.0, + 0.0, + 0.0, + -0.028147941798344255, + 0.0, + 0.0 + ], + "q99": [ + 0.04998000074476004, + 0.04998000074476004, + 0.04998000074476004, + 0.0, + 0.0, + 0.0, + 0.08601800156980753, + 0.3782101800918579, + 0.3667341671466828, + 0.3667019561767578, + 0.04438944834824652, + 0.36664280390143394, + 0.3667250802397728, + 0.36671328068375586, + 0.0382236413366627, + 0.3666137705862522, + 0.3667169698476791, + 0.3667099578022957, + 0.2819846283018589, + 0.1071048165038228, + 0.36666986070871355, + 0.3667560145735741, + 0.04998000074476004, + 0.04998000074476004, + 0.021540000320971017, + 0.0, + 0.0, + 0.0, + 0.011421882868045942, + 0.36662054525613785, + 0.36667089745402337, + 0.3667560145735741, + 0.0020719100366113708, + 0.36662088149786, + 0.3666709254741669, + 0.3667560145735741, + 0.0010109457788770669, + 0.36662124575972554, + 0.3666709815144539, + 0.3667560145735741, + 0.2821533048152924, + 0.005419915117323398, + 0.3665988940000534, + 0.36667801457047466 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmOpenPen-v1/source_audit.json b/previous_single_task/DualArmOpenPen-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..2edf1865d57d15cbf57d2f4ebe13da15db46df35 --- /dev/null +++ b/previous_single_task/DualArmOpenPen-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmOpenPen-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.65", + "peak_step": "30000", + "final": "0.35", + "start": "2026-08-22 05:49:01", + "eval_mtime": "2026-08-25 06:05:33.818434", + "source_mtime": "2026-06-21 06:12:26.475111", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_dex/level_1/open_pen.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmOpenPen-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmOpenPen-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmOpenPen-v1/lora_params/lora_step_0030000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmPegInHole-v1/eval_result.json b/previous_single_task/DualArmPegInHole-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..04dbe7cab05f4f3879ca64f995f4026f56061be0 --- /dev/null +++ b/previous_single_task/DualArmPegInHole-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmPegInHole-v1", + "level": "level_2", + "suite": "rigid_body/dual_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 9, + "success_rate": 0.18, + "return_mean": 84.89753922495059, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmPegInHole-v1/evaluation_history.csv b/previous_single_task/DualArmPegInHole-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..1adf286f48ba8b452b39d70fc667cc41b719d0ae --- /dev/null +++ b/previous_single_task/DualArmPegInHole-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,9,0.18,84.89753922495059,True +10000,50,6,0.12,61.962898916585836,True +15000,50,5,0.1,51.43035095642437,True +20000,50,4,0.08,37.64026053449139,True +25000,50,5,0.1,34.97058293002657,True +30000,50,7,0.14,55.99565888670273,True +35000,50,6,0.12,41.83745249252534,True +40000,50,8,0.16,52.70686611412093,True diff --git a/previous_single_task/DualArmPegInHole-v1/lora_step_0005000.npz b/previous_single_task/DualArmPegInHole-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..c1c868f65f8255a08a11e669ca3b39328b5e1e0a --- /dev/null +++ b/previous_single_task/DualArmPegInHole-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:3da1d0202466bf4a0182a47ddb00ecd42112a22ed462c85197a9b5625f4d5508 +size 1867844866 diff --git a/previous_single_task/DualArmPegInHole-v1/metadata.json b/previous_single_task/DualArmPegInHole-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..7eba0b8fb57a48f24f15b6e7338fb5f1ea46d8ba --- /dev/null +++ b/previous_single_task/DualArmPegInHole-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmPegInHole-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "3da1d0202466bf4a0182a47ddb00ecd42112a22ed462c85197a9b5625f4d5508", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "fea9f657e409e787be3bd64e284b1b7535c3e08c435f247c5dfc030e19f48a5a", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmPegInHole-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_gripper/level_2/peg_in_hole.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmPegInHole-v1/norm_stats.json b/previous_single_task/DualArmPegInHole-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..305828a1264ef1d9bfb50f7c25d6245976861e2a --- /dev/null +++ b/previous_single_task/DualArmPegInHole-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0015893884701654315, + -0.009668529964983463, + -0.009714990854263306, + -0.0003838927368633449, + -2.8560609734995523e-6, + 0.011924734339118004, + -0.4766727089881897, + 0.0031141263898462057, + 0.015353817492723465, + -0.010168387554585934, + 0.0008654639241285622, + -0.000018920582078862935, + -0.006882473360747099, + -0.4229753017425537, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.043559730052948, + 0.03892677277326584, + 0.05185994878411293, + 0.004467052407562733, + 0.0006886181072331965, + 0.060708943754434586, + 0.8790808320045471, + 0.039488278329372406, + 0.04660831019282341, + 0.05910274013876915, + 0.006370931398123503, + 0.0004203076532576233, + 0.05272228643298149, + 0.9061412215232849, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + -0.0093334519304335, + -0.002037983856722712, + -0.09999989718198776, + -1.0, + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + -0.0010344516462646425, + -0.0014285594180226328, + -0.17937153596878053, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0008634385466575745, + 0.0017934497762471437, + 0.1757442134037614, + 0.9996, + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.03378593628352974, + 0.0011725386682897802, + 0.09994364809840917, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmPegInHole-v1/source_audit.json b/previous_single_task/DualArmPegInHole-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..369a8a18ba96cf0b7175ccacf1e11bcc7d439799 --- /dev/null +++ b/previous_single_task/DualArmPegInHole-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmPegInHole-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.25", + "peak_step": "5000", + "final": "0.05", + "start": "2026-08-22 06:11:16", + "eval_mtime": "2026-08-25 03:18:57.813853", + "source_mtime": "2026-05-02 05:38:20.461656", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_gripper/level_2/peg_in_hole.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmPegInHole-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmPegInHole-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmPegInHole-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmPingpongBalance-v1/eval_result.json b/previous_single_task/DualArmPingpongBalance-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..b619453fa86f9852383adc5fd22da1f8ac00dd86 --- /dev/null +++ b/previous_single_task/DualArmPingpongBalance-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmPingpongBalance-v1", + "level": "level_3", + "suite": "rigid_body/dual_arm_dex", + "step": 15000, + "episodes": 50, + "successes": 10, + "success_rate": 0.2, + "return_mean": 762.3643163096905, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmPingpongBalance-v1/evaluation_history.csv b/previous_single_task/DualArmPingpongBalance-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..9af856a398c115fb0739766edce7886f1d4e209f --- /dev/null +++ b/previous_single_task/DualArmPingpongBalance-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,1,0.02,120.94932576864957,True +10000,50,1,0.02,558.1151409831643,True +15000,50,10,0.2,762.3643163096905,True +20000,50,3,0.06,827.9384214293957,True +25000,50,2,0.04,855.3278978893161,True +30000,50,3,0.06,936.852864215076,True +35000,50,5,0.1,918.2753744417429,True +40000,50,7,0.14,845.8957908791303,True diff --git a/previous_single_task/DualArmPingpongBalance-v1/lora_step_0015000.npz b/previous_single_task/DualArmPingpongBalance-v1/lora_step_0015000.npz new file mode 100644 index 0000000000000000000000000000000000000000..bfe913cb55cf9c44531a3644dd3204834488feec --- /dev/null +++ b/previous_single_task/DualArmPingpongBalance-v1/lora_step_0015000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:da1fb459a8bcc79cb0985cb6e78eb4e477f7c2a88908422063379612734dab5d +size 1867943218 diff --git a/previous_single_task/DualArmPingpongBalance-v1/metadata.json b/previous_single_task/DualArmPingpongBalance-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..63cb118d1662a877ec36f312d4cfb9c068be7176 --- /dev/null +++ b/previous_single_task/DualArmPingpongBalance-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmPingpongBalance-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0015000.npz", + "checkpoint_sha256": "da1fb459a8bcc79cb0985cb6e78eb4e477f7c2a88908422063379612734dab5d", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 15000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "4983637c5b1f06148d7aaba61e537379bbd3baf11362050f5acc3396658ca8fd", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmPingpongBalance-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_dex/level_3/pingpong_balance.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmPingpongBalance-v1/norm_stats.json b/previous_single_task/DualArmPingpongBalance-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..c1927e7c8b2fd7342317343c6002b4c6ccf2949c --- /dev/null +++ b/previous_single_task/DualArmPingpongBalance-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.0030920696444809437, + -0.0066130138002336025, + 0.006300327368080616, + 0.0, + 0.0, + 0.0, + -0.001735094701871276, + 0.011512120254337788, + 0.008346614427864552, + 0.004974243696779013, + -0.0003489268710836768, + 0.0024182626511901617, + 0.0010036291787400842, + 0.00036125624319538474, + -0.0002784997341223061, + -0.0005361207877285779, + -0.0003420180582907051, + -0.0004552266327664256, + -0.027942344546318054, + -0.0004225240263622254, + 0.0033697665203362703, + -0.00005424420305644162, + -0.0034801396541297436, + 0.00858578085899353, + -0.013812482357025146, + 0.00003869592910632491, + 0.004664209671318531, + 0.0, + -0.000037549645639956, + 0.03643405809998512, + 0.034358225762844086, + 0.03409016504883766, + 0.000020133787984377705, + 0.036501042544841766, + 0.03746798262000084, + 0.0360676571726799, + 0.00003048533653782215, + 0.03571724891662598, + 0.03657984361052513, + 0.03574616461992264, + 0.23874004185199738, + 0.000214176150620915, + 0.14645835757255554, + 0.11140413582324982 + ], + "std": [ + 0.01836310513317585, + 0.03216119483113289, + 0.034956227988004684, + 0.0, + 0.0, + 0.0, + 0.004877972882241011, + 0.1699630618095398, + 0.08730018883943558, + 0.03892814368009567, + 0.0009243901004083455, + 0.16269971430301666, + 0.0023546896409243345, + 0.029139185324311256, + 0.00038193329237401485, + 0.16228345036506653, + 0.08116961270570755, + 0.08700688183307648, + 0.16238677501678467, + 0.0007659480907022953, + 0.0222293920814991, + 0.014796710573136806, + 0.019781244918704033, + 0.023748096078634262, + 0.028098145499825478, + 0.0028948807157576084, + 0.012965168803930283, + 0.0, + 0.00010979310900438577, + 0.10152766108512878, + 0.10142016410827637, + 0.10111250728368759, + 0.00017885035776998848, + 0.10138212144374847, + 0.10050534456968307, + 0.10051940381526947, + 0.00017455463239457458, + 0.10152112692594528, + 0.10067357122898102, + 0.10058921575546265, + 0.14599187672138214, + 0.0007826769142411649, + 0.10668676346540451, + 0.09712676703929901 + ], + "q01": [ + -0.05000000074505806, + -0.05000000074505806, + -0.05000000074505806, + 0.0, + 0.0, + 0.0, + -0.022695185492638575, + -0.5991425752639771, + -0.2995628261566162, + -0.12838027322292328, + -0.004797642931766859, + -0.5986405482292175, + -0.000012627931027964223, + -0.10699379110336305, + -0.0015583088886371116, + -0.5982857646942139, + -0.2993234009742737, + -0.32099023723602294, + -0.7907933111667633, + -0.001940016758395359, + -0.08911194971650839, + -0.06418998396396637, + -0.05000000074505806, + -0.05000000074505806, + -0.05000000074505806, + -0.007300584428012371, + -0.010955838644877076, + 0.0, + -0.0005917208745027892, + -0.000019153305402141996, + -2.3768041046423605e-6, + 0.0, + -0.0007366570039419458, + -0.00001854182846727781, + -2.300177584402263e-6, + 0.0, + -0.0006933948825928383, + -0.00001792975490388926, + -2.222936473117443e-6, + 0.0, + -2.3625225367140956e-6, + -0.0005622803000733256, + -1.1339667480569915e-6, + -1.3726678460557196e-8 + ], + "q99": [ + 0.04998000074476004, + 0.04998000074476004, + 0.04998000074476004, + 0.0, + 0.0, + 0.0, + 1.6978326454579473e-6, + 0.5988791087269782, + 0.29945719757676126, + 0.12834152653217318, + 5.320895020759393e-6, + 0.5986397822380065, + 0.01172542349621508, + 0.10695048306286334, + 6.08036450412317e-6, + 0.5986689452409746, + 0.29947621003985403, + 0.32085075955986975, + 0.32733388917446127, + 0.0011168490735813975, + 0.04322368731945753, + 0.029223296067118654, + 0.04998000074476004, + 0.04998000074476004, + 0.04998000074476004, + 0.007189560525119303, + 0.05860820133425294, + 0.0, + 0.00006171775473922021, + 0.5561207415955017, + 0.5561376474787649, + 0.5561399796009063, + 0.00024026076325681057, + 0.5561212888723778, + 0.5561376585436345, + 0.5561399796009063, + 0.00020446438532089814, + 0.556121887233141, + 0.5561377716927192, + 0.5561399796009063, + 0.4278198539544374, + 0.002176426012767479, + 0.5561397138606399, + 0.556139977618774 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmPingpongBalance-v1/source_audit.json b/previous_single_task/DualArmPingpongBalance-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..fbd8c13c3024b1520a2b2a91e2b7be011b68849c --- /dev/null +++ b/previous_single_task/DualArmPingpongBalance-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmPingpongBalance-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.2", + "peak_step": "20000", + "final": "0.2", + "start": "2026-08-22 06:01:08", + "eval_mtime": "2026-08-25 05:12:16.815218", + "source_mtime": "2026-05-27 12:52:11.672733", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_dex/level_3/pingpong_balance.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmPingpongBalance-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmPingpongBalance-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmPingpongBalance-v1/lora_params/lora_step_0020000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmPlaceRopeOverRod-v1/eval_result.json b/previous_single_task/DualArmPlaceRopeOverRod-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..911112c8aa533492d8a5f0c57da12b23da42a227 --- /dev/null +++ b/previous_single_task/DualArmPlaceRopeOverRod-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmPlaceRopeOverRod-v1", + "level": "level_2", + "suite": "deformable/dual_arm_dex", + "step": 40000, + "episodes": 50, + "successes": 23, + "success_rate": 0.46, + "return_mean": 289.2976510642795, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmPlaceRopeOverRod-v1/evaluation_history.csv b/previous_single_task/DualArmPlaceRopeOverRod-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..b733cf22520755e749d539bbf26cf7afd9f6181e --- /dev/null +++ b/previous_single_task/DualArmPlaceRopeOverRod-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,1,0.02,159.39394753875212,True +10000,50,2,0.04,196.88293693638406,True +15000,50,7,0.14,289.13082848131654,True +20000,50,17,0.34,243.5225198026281,True +25000,50,14,0.28,324.9540537391882,True +30000,50,19,0.38,296.7787297666073,True +35000,50,4,0.08,298.22417598265224,True +40000,50,23,0.46,289.2976510642795,True diff --git a/previous_single_task/DualArmPlaceRopeOverRod-v1/lora_step_0040000.npz b/previous_single_task/DualArmPlaceRopeOverRod-v1/lora_step_0040000.npz new file mode 100644 index 0000000000000000000000000000000000000000..7c94e524c63b0f39d9937171f740855db8c5e673 --- /dev/null +++ b/previous_single_task/DualArmPlaceRopeOverRod-v1/lora_step_0040000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ab110fb81eabcc3bd39bb226cd5856d3b8be169e6ec2745d0b8aebaa94d2ac5e +size 1867943218 diff --git a/previous_single_task/DualArmPlaceRopeOverRod-v1/metadata.json b/previous_single_task/DualArmPlaceRopeOverRod-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..fa23ae39d54b01decb9ab1ae3bc282a79d2859c5 --- /dev/null +++ b/previous_single_task/DualArmPlaceRopeOverRod-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmPlaceRopeOverRod-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0040000.npz", + "checkpoint_sha256": "ab110fb81eabcc3bd39bb226cd5856d3b8be169e6ec2745d0b8aebaa94d2ac5e", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 40000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "71519bece4284989edc29c37bef8dd9cba44aa2805e982727febd3c9d8417533", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmPlaceRopeOverRod-v1", + "source_path": "maniladder/envs/deformable/dual_arm_dex/level_2/drape_rope.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmPlaceRopeOverRod-v1/norm_stats.json b/previous_single_task/DualArmPlaceRopeOverRod-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..cf32ce84ef529eef638d39f39e9b2db29356f151 --- /dev/null +++ b/previous_single_task/DualArmPlaceRopeOverRod-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.006324111483991146, + 0.005130593664944172, + -0.004154087509959936, + 0.0, + 0.0, + 0.0, + -0.00048530386993661523, + 0.010310346260666847, + 0.004393214359879494, + 0.0012032095110043883, + 0.0003606509417295456, + -0.0026184769812971354, + -0.0008463854901492596, + 0.0012414039811119437, + 0.0014517739182338119, + -0.008939112536609173, + -0.006828959099948406, + -0.004143157508224249, + 0.0016950236167758703, + -0.00031181215308606625, + 0.0005322943325154483, + 0.0013105662073940039, + -0.0039162407629191875, + -0.02048657089471817, + -0.004904493689537048, + 0.0, + 0.0, + 0.0, + 0.0004624160937964916, + 0.03526610508561134, + 0.019378403201699257, + 0.021128464490175247, + -0.000024108718207571656, + 0.029225420206785202, + 0.016936471685767174, + 0.02138679474592209, + -0.00011523016291903332, + 0.02918383479118347, + 0.017076119780540466, + 0.021427543833851814, + -0.02306811511516571, + -0.001713556470349431, + 0.005264640785753727, + 0.027353521436452866 + ], + "std": [ + 0.04393837973475456, + 0.055775392800569534, + 0.03672945499420166, + 0.0, + 0.0, + 0.0, + 0.0006281888927333057, + 0.1199551522731781, + 0.07144352793693542, + 0.08929703384637833, + 0.001117074047215283, + 0.125543475151062, + 0.07391165941953659, + 0.08895262330770493, + 0.0031341994181275368, + 0.13635599613189697, + 0.08416862785816193, + 0.09711414575576782, + 0.062480852007865906, + 0.00045652512926608324, + 0.0007409352110698819, + 0.1030997633934021, + 0.021224776282906532, + 0.041872721165418625, + 0.027617881074547768, + 0.0, + 0.0, + 0.0, + 0.0004734526155516505, + 0.07625654339790344, + 0.04576599970459938, + 0.05886431783437729, + 0.00006530334212584421, + 0.07751838862895966, + 0.046528179198503494, + 0.0587698295712471, + 0.00035885945544578135, + 0.07705599069595337, + 0.04647420719265938, + 0.05875536799430847, + 0.046899281442165375, + 0.00447303568944335, + 0.010405784472823143, + 0.0704590305685997 + ], + "q01": [ + -0.14000000059604645, + -0.10000000149011612, + -0.05508799985796213, + 0.0, + 0.0, + 0.0, + -0.001985242194030434, + -0.4036341611683369, + -0.2412840978384018, + -0.30911775538921354, + -0.0001651560085825622, + -0.41133638626337055, + -0.2441774514079094, + -0.3084424785614014, + -0.00002536176994908601, + -0.434119999563694, + -0.2512096124887466, + -0.31370662751197814, + -0.21627607090473175, + -0.0012663842790294438, + -0.0012849821679294109, + -0.3586417397916317, + -0.14000000059604645, + -0.14000000059604645, + -0.055034000234305866, + 0.0, + 0.0, + 0.0, + -0.00030459497240372004, + -0.000016224894352490082, + -1.8720443222264294e-6, + -0.0007174015045166016, + -0.0003143666246032808, + -0.000033736228942871094, + -1.8000132513407152e-6, + -8.502295756329659e-9, + -0.0018617678687151058, + -0.000015076959243742749, + -1.7275106074521318e-6, + -8.503636905743406e-9, + -0.21846472542967532, + -0.022958563604368827, + -0.00002509876503609121, + -0.0002569464564323425 + ], + "q99": [ + 0.07995599821209906, + 0.1399520005956292, + 0.09995200148969888, + 0.0, + 0.0, + 0.0, + 0.0012814806830137963, + 0.3998257952153682, + 0.24397398523092273, + 0.30839636037349705, + 0.005821360527910292, + 0.39982401108145715, + 0.24407272269129754, + 0.3083044402122497, + 0.015307315189740621, + 0.3998200059592725, + 0.24404067235589033, + 0.3082982861280441, + 0.21840647888183595, + 0.00026955668712034826, + 0.0023201968491077425, + 0.36051983737349513, + 0.022559934625402095, + 0.0599599986590445, + 0.06260800026655194, + 0.0, + 0.0, + 0.0, + 0.0009725281407940202, + 0.3999200027142935, + 0.24424327341914476, + 0.30834528036117553, + 0.000025150500448944534, + 0.3999199992120266, + 0.2441293347378448, + 0.30837601103453205, + 0.000024915906820387797, + 0.3999200029438805, + 0.24407242168588927, + 0.30837601103433976, + -0.00003108591880734979, + 0.00046022883432451575, + 0.048302666538977065, + 0.36051887800097465 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmPlaceRopeOverRod-v1/source_audit.json b/previous_single_task/DualArmPlaceRopeOverRod-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..a8b110b6ef669d7e5035bf00cb61492a3c1a3811 --- /dev/null +++ b/previous_single_task/DualArmPlaceRopeOverRod-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmPlaceRopeOverRod-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.75", + "peak_step": "30000", + "final": "0.7", + "start": "2026-09-04 06:24:17", + "eval_mtime": "2026-09-07 19:11:00.381252", + "source_mtime": "2026-08-17 07:45:52.155829", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_dex/level_2/drape_rope.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmPlaceRopeOverRod-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmPlaceRopeOverRod-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmPlaceRopeOverRod-v1/lora_params/lora_step_0030000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmPouringWater-v1/eval_result.json b/previous_single_task/DualArmPouringWater-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..ae29ff322ad63321f4844e0281ef31794ed77d04 --- /dev/null +++ b/previous_single_task/DualArmPouringWater-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmPouringWater-v1", + "level": "level_1", + "suite": "deformable/dual_arm_gripper", + "step": 30000, + "episodes": 20, + "successes": 19, + "success_rate": 0.95, + "return_mean": 153.65338471363066, + "return_mean_finite": true, + "episode_seed_base": null, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "historical_online_eval20", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmPouringWater-v1/evaluation_history.csv b/previous_single_task/DualArmPouringWater-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..2ddc5ab78c6d434ce363b82aef4f331340335e32 --- /dev/null +++ b/previous_single_task/DualArmPouringWater-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,20,16,0.8,142.2999885714904,True +10000,20,12,0.6,145.42681437809952,True +15000,20,13,0.65,128.7162898265291,True +20000,20,16,0.8,146.2656669145159,True +25000,20,16,0.8,141.91980147731957,True +30000,20,19,0.95,153.65338471363066,True +35000,20,17,0.85,153.00548436687677,True +40000,20,17,0.85,158.49867949652835,True diff --git a/previous_single_task/DualArmPouringWater-v1/metadata.json b/previous_single_task/DualArmPouringWater-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..6cf3440f916fe80692410a4de7edc77e573eb3bf --- /dev/null +++ b/previous_single_task/DualArmPouringWater-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmPouringWater-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0030000.npz", + "checkpoint_sha256": "d1a0affb3e58bc3eb0b23f5ae83cb9a7f64a8be66940868409a87cb0b1b4ca34", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 30000, + "eval_episodes": 20, + "eval_seed_base": null, + "norm_stats_sha256": "cf86c0955d52f5f0555b9ddc7bd21f02e075895b692fe962cbafd5315fda674a", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmPouringWater-v1", + "source_path": "maniladder/envs/deformable/dual_arm_gripper/level_1/pouring_water.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmPouringWater-v1/norm_stats.json b/previous_single_task/DualArmPouringWater-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..652d48a68ee6bb9dec8cd902e19eaaeaa07ec7f6 --- /dev/null +++ b/previous_single_task/DualArmPouringWater-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.030269382521510124, + 0.005403217859566212, + -0.0589505210518837, + -0.012078359723091125, + -0.05715694651007652, + 0.011172072030603886, + -0.8661856055259705, + -0.029509153217077255, + -0.008053317666053772, + -0.06384208798408508, + 0.014146174304187298, + -0.05816637724637985, + -0.010399560444056988, + -0.8661856055259705, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.14164075255393982, + 0.0324835442006588, + 0.13165588676929474, + 0.027065617963671684, + 0.05177653208374977, + 0.03020567074418068, + 0.4997223913669586, + 0.13712330162525177, + 0.0348350815474987, + 0.13495023548603058, + 0.030000831931829453, + 0.05260077863931656, + 0.02639252319931984, + 0.4997223913669586, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.3860348678231239, + -0.10093492822647096, + -0.37252119815349577, + -0.0947582758128643, + -0.16674889355478809, + -0.004012787877023222, + -1.0, + -0.37179443497657777, + -0.10586176637411118, + -0.37590238672494886, + -0.005735298663377762, + -0.15909418766573072, + -0.11588376990407705, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.35763335436582566, + 0.0947834491729736, + 0.29429272317886357, + 0.005734613675996661, + 0.004640274342615158, + 0.13110186223089695, + 0.9996, + 0.3576228357791901, + 0.10554187929034231, + 0.29430713140964504, + 0.10303689595460892, + 0.0045228373672812905, + 0.0037299621611833478, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmPouringWater-v1/source_audit.json b/previous_single_task/DualArmPouringWater-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..921bd650e72b95466a8db952fd6b6bf6b6c8e80a --- /dev/null +++ b/previous_single_task/DualArmPouringWater-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmPouringWater-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.95", + "peak_step": "30000", + "final": "0.85", + "start": "2026-09-04 06:04:15", + "eval_mtime": "2026-09-07 19:30:59.957704", + "source_mtime": "2026-08-29 08:02:41.881800", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_gripper/level_1/pouring_water.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmPouringWater-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmPouringWater-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmPouringWater-v1/lora_params/lora_step_0030000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmRopeCylinderWarpping-v1/eval_result.json b/previous_single_task/DualArmRopeCylinderWarpping-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..04ef7b1876d62d43e53f182f1a60abc93c4a71f0 --- /dev/null +++ b/previous_single_task/DualArmRopeCylinderWarpping-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmRopeCylinderWarpping-v1", + "level": "level_1", + "suite": "deformable/dual_arm_gripper", + "step": 10000, + "episodes": 50, + "successes": 31, + "success_rate": 0.62, + "return_mean": 81.67051498234272, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmRopeCylinderWarpping-v1/evaluation_history.csv b/previous_single_task/DualArmRopeCylinderWarpping-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..39f8fb780a27854ac19847268df10067ff1db9e8 --- /dev/null +++ b/previous_single_task/DualArmRopeCylinderWarpping-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,28,0.56,88.58144970148801,True +10000,50,31,0.62,81.67051498234272,True +15000,50,27,0.54,86.22807513743639,True +20000,50,24,0.48,85.54877895057201,True +25000,50,25,0.5,88.75803962796927,True +30000,50,27,0.54,86.46499432876706,True +35000,50,26,0.52,83.81079191997647,True +40000,50,25,0.5,88.3807731771469,True diff --git a/previous_single_task/DualArmRopeCylinderWarpping-v1/lora_step_0010000.npz b/previous_single_task/DualArmRopeCylinderWarpping-v1/lora_step_0010000.npz new file mode 100644 index 0000000000000000000000000000000000000000..107a863f6fef990e1b73548295863d4a61eaa524 --- /dev/null +++ b/previous_single_task/DualArmRopeCylinderWarpping-v1/lora_step_0010000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a75996da6ae851c12335001a14a54ce38cf3038772fc242de85651ecd4992c6a +size 1867844866 diff --git a/previous_single_task/DualArmRopeCylinderWarpping-v1/metadata.json b/previous_single_task/DualArmRopeCylinderWarpping-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..78212a7ae0036669c854d36c90d4749633e0df33 --- /dev/null +++ b/previous_single_task/DualArmRopeCylinderWarpping-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmRopeCylinderWarpping-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0010000.npz", + "checkpoint_sha256": "a75996da6ae851c12335001a14a54ce38cf3038772fc242de85651ecd4992c6a", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 10000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "c0bc915def6aa6e9d129e1bfa49eeeec15ab037749a0bdb35ee9e698f74ca9fa", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmRopeCylinderWarpping-v1", + "source_path": "maniladder/envs/deformable/dual_arm_gripper/level_1/rope_cylinder_wrapping.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmRopeCylinderWarpping-v1/norm_stats.json b/previous_single_task/DualArmRopeCylinderWarpping-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..bd5d63916b6147ee131739fbf03d54418ef9de16 --- /dev/null +++ b/previous_single_task/DualArmRopeCylinderWarpping-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.01454230211675167, + -0.0033801875542849302, + -0.03720851615071297, + 0.0, + 0.0, + 0.00024738433421589434, + -0.4137105941772461, + 0.01117957103997469, + -0.0034748241305351257, + -0.03684453293681145, + 0.0, + 0.0, + 0.00010645806469256058, + -0.4137105941772461, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.06602320820093155, + 0.04471374675631523, + 0.04517922177910805, + 0.0, + 0.0, + 0.004614241886883974, + 0.9104084372520447, + 0.06482572108507156, + 0.05084962025284767, + 0.044707585126161575, + 0.0, + 0.0, + 0.004900801926851273, + 0.9104084372520447, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10002625632286072, + -0.10000800397396087, + -0.13500000536441803, + 0.0, + 0.0, + 0.0, + -1.0, + -0.10000800397396087, + -0.13500000536441803, + -0.13500000536441803, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.13494818038940432, + 0.13494600536227225, + 0.09997568206489085, + 0.0, + 0.0, + 0.049100000731647025, + 0.9996, + 0.13494600536227225, + 0.09995400397181511, + 0.09995300148874522, + 0.0, + 0.0, + 0.015600000232458111, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmRopeCylinderWarpping-v1/source_audit.json b/previous_single_task/DualArmRopeCylinderWarpping-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..e2de010ade92742a7b9dd7cf4c00311478eee9bb --- /dev/null +++ b/previous_single_task/DualArmRopeCylinderWarpping-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmRopeCylinderWarpping-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.7", + "peak_step": "30000", + "final": "0.6", + "start": "2026-08-22 06:01:11", + "eval_mtime": "2026-08-25 07:48:44.066731", + "source_mtime": "2026-08-09 12:24:54.333809", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_gripper/level_1/rope_cylinder_wrapping.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmRopeCylinderWarpping-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmRopeCylinderWarpping-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmRopeCylinderWarpping-v1/lora_params/lora_step_0030000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmRopeStretch-v1/eval_result.json b/previous_single_task/DualArmRopeStretch-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..d0c884d2cc2cbbefd7563c4cc698a78e1150a0da --- /dev/null +++ b/previous_single_task/DualArmRopeStretch-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmRopeStretch-v1", + "level": "level_1", + "suite": "deformable/dual_arm_dex", + "step": 10000, + "episodes": 50, + "successes": 46, + "success_rate": 0.92, + "return_mean": 186.05761841654777, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmRopeStretch-v1/evaluation_history.csv b/previous_single_task/DualArmRopeStretch-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..f703758725831714ba60a641a238325d6bba7430 --- /dev/null +++ b/previous_single_task/DualArmRopeStretch-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,30,0.6,217.4085968887806,True +10000,50,46,0.92,186.05761841654777,True +15000,50,46,0.92,183.99752925515176,True +20000,50,41,0.82,195.065522377491,True +25000,50,41,0.82,195.7638722229004,True +30000,50,43,0.86,191.87583116173744,True +35000,50,38,0.76,202.07684795379637,True +40000,50,38,0.76,201.48189628362655,True diff --git a/previous_single_task/DualArmRopeStretch-v1/lora_step_0010000.npz b/previous_single_task/DualArmRopeStretch-v1/lora_step_0010000.npz new file mode 100644 index 0000000000000000000000000000000000000000..febde1c76419fd79c9f63e270d558e549f7529fd --- /dev/null +++ b/previous_single_task/DualArmRopeStretch-v1/lora_step_0010000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:50616181d63062ba47a5a893add02f9b6a99743801d4690f62f11ecb1250feda +size 1867943218 diff --git a/previous_single_task/DualArmRopeStretch-v1/metadata.json b/previous_single_task/DualArmRopeStretch-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..aa01d5bcf6a82421a761f413ff5bc334c385a7ae --- /dev/null +++ b/previous_single_task/DualArmRopeStretch-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmRopeStretch-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0010000.npz", + "checkpoint_sha256": "50616181d63062ba47a5a893add02f9b6a99743801d4690f62f11ecb1250feda", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 10000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "8cd769f2f84d30c9d378b19663107bdd506d05b6b0bfd9398477c416993bfd52", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmRopeStretch-v1", + "source_path": "maniladder/envs/deformable/dual_arm_dex/level_1/rope_stretch.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmRopeStretch-v1/norm_stats.json b/previous_single_task/DualArmRopeStretch-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..57eb6a5448d5b81ce92fb9fce6de488aff2e8c30 --- /dev/null +++ b/previous_single_task/DualArmRopeStretch-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.0008398302015848458, + -0.0007015590672381222, + 0.03241126611828804, + 0.0, + 0.0, + 0.0, + -0.0010093202581629157, + 0.09411652386188507, + 0.06486862897872925, + 0.06290509551763535, + -8.410857503804436e-7, + 0.061812181025743484, + 0.06182136759161949, + 0.06182130053639412, + 0.00003121925692539662, + 0.06193140149116516, + 0.061865199357271194, + 0.06182428449392319, + 0.05412151664495468, + 0.0071080587804317474, + 0.12358305603265762, + 0.08212016522884369, + -0.0006841711583547294, + 0.0009727090946398675, + 0.03239854425191879, + 0.0, + 0.0, + 0.0, + 0.00010770249355118722, + 0.06280152499675751, + 0.0619855672121048, + 0.06187988072633743, + 2.5246990844607353e-6, + 0.061813559383153915, + 0.061827950179576874, + 0.06182851642370224, + -0.00003117335290880874, + 0.06193133071064949, + 0.061911020427942276, + 0.061849143356084824, + 0.17746995389461517, + 0.00036179396556690335, + 0.09954685717821121, + 0.08765251934528351 + ], + "std": [ + 0.01097572036087513, + 0.005565279629081488, + 0.05783875659108162, + 0.0, + 0.0, + 0.0, + 0.0024155485443770885, + 0.07996536791324615, + 0.08631275594234467, + 0.08716335147619247, + 3.6687386000266997e-6, + 0.08763490617275238, + 0.08764367550611496, + 0.0876460149884224, + 0.000029922439352958463, + 0.08768235892057419, + 0.08763977140188217, + 0.0876430794596672, + 0.11409033834934235, + 0.01221404504030943, + 0.08201275020837784, + 0.08112754672765732, + 0.01021274458616972, + 0.0067038824781775475, + 0.057846661657094955, + 0.0, + 0.0, + 0.0, + 0.00012797056115232408, + 0.0872340202331543, + 0.08758869022130966, + 0.08762887120246887, + 1.575617943672114e-6, + 0.08763666450977325, + 0.08764491230249405, + 0.08764602243900299, + 0.00003531378388288431, + 0.0876723974943161, + 0.08762723952531815, + 0.08763288706541061, + 0.09630057960748672, + 0.00037204602267593145, + 0.07854525744915009, + 0.07930130511522293 + ], + "q01": [ + -0.07999999821186066, + -0.012415999722480767, + -0.07999999821186066, + 0.0, + 0.0, + 0.0, + -0.004777193327434361, + -0.000012589073776325677, + -1.5542954088232364e-6, + -2.096055107259076e-9, + -0.000014594220918115752, + -0.000012070350749127101, + -1.4882463119647582e-6, + -2.095388307310486e-9, + -2.9510987471439874e-6, + -0.000011550883755262475, + -1.421781121280219e-6, + -2.0941786083028546e-9, + -0.000014333681065181736, + -0.00009666685353731736, + -7.965973964019213e-6, + -6.508562222506953e-8, + -0.04262399904727936, + -0.0005256851017475146, + -0.07999999821186066, + 0.0, + 0.0, + 0.0, + -1.2927939115797925e-6, + -0.000012513030924310442, + -1.5188189763648552e-6, + -2.1566706198683505e-9, + -1.150636679813033e-6, + -0.000012010461432510056, + -1.4553086202795384e-6, + -2.1556791907073602e-9, + -0.00012632351171996563, + -0.000011507536328281276, + -1.3914147984905867e-6, + -2.154539213705675e-9, + -0.000012114442142774351, + -3.956386227946496e-6, + -7.968017598614097e-6, + -6.662722995542936e-8 + ], + "q99": [ + 0.016447999632358545, + 0.000639999985694889, + 0.07996799821257591, + 0.0, + 0.0, + 0.0, + 0.005362922426313163, + 0.3639097592209624, + 0.36392498736886086, + 0.36392719261008305, + 4.08796161518694e-6, + 0.3639104148446693, + 0.3639250767711564, + 0.36392719261008316, + 0.00009081513936107512, + 0.36391107046852506, + 0.36392516617353526, + 0.3639271926100834, + 0.3999200030925362, + 0.04562865272657364, + 0.36393148169342443, + 0.36392725219020894, + 0.01375999969244003, + 0.024658857464790344, + 0.07996799821257591, + 0.0, + 0.0, + 0.0, + 0.00041005458415620526, + 0.3637642207765404, + 0.36392498737595613, + 0.36392719261007095, + 5.168608816165942e-6, + 0.36376487633578397, + 0.36385230670175484, + 0.3639271926100711, + 4.063686590961878e-6, + 0.363765472326359, + 0.36385236630811535, + 0.36392719261007134, + 0.39992000353638396, + 0.0008704503609878884, + 0.36385888777539366, + 0.3638544521661325 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmRopeStretch-v1/source_audit.json b/previous_single_task/DualArmRopeStretch-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..fa90b0ee11ae9191ef12238a6c65e80c7289127b --- /dev/null +++ b/previous_single_task/DualArmRopeStretch-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmRopeStretch-v1", + "category": "reuse", + "level": "level_1", + "peak": "1.0", + "peak_step": "10000", + "final": "1.0", + "start": "2026-08-21 13:36:52", + "eval_mtime": "2026-08-25 11:50:13.484175", + "source_mtime": "2026-08-12 05:02:15.394823", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_dex/level_1/rope_stretch.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmRopeStretch-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmRopeStretch-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmRopeStretch-v1/lora_params/lora_step_0010000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmSendPingpong-v1/eval_result.json b/previous_single_task/DualArmSendPingpong-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..a16ba6aa517c5ac801a87a9c97e532a6cb6b118d --- /dev/null +++ b/previous_single_task/DualArmSendPingpong-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmSendPingpong-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_dex", + "step": 40000, + "episodes": 50, + "successes": 4, + "success_rate": 0.08, + "return_mean": 476.9731384050846, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmSendPingpong-v1/evaluation_history.csv b/previous_single_task/DualArmSendPingpong-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..5bff8d96263f437cb930cdff7a8d4ec83668806e --- /dev/null +++ b/previous_single_task/DualArmSendPingpong-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,493.29558154344556,True +10000,50,1,0.02,437.45649006843564,True +15000,50,0,0.0,556.9836190354824,True +20000,50,0,0.0,484.0609417641163,True +25000,50,0,0.0,494.965841627121,True +30000,50,0,0.0,477.9118604850769,True +35000,50,3,0.06,481.6546700167656,True +40000,50,4,0.08,476.9731384050846,True diff --git a/previous_single_task/DualArmSendPingpong-v1/lora_step_0040000.npz b/previous_single_task/DualArmSendPingpong-v1/lora_step_0040000.npz new file mode 100644 index 0000000000000000000000000000000000000000..c22e25c26e3c4115424296da986fdffcfc955d78 --- /dev/null +++ b/previous_single_task/DualArmSendPingpong-v1/lora_step_0040000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:70924c6c7880fc34d61197d289cbbbbdc15a3e146339aed6b10af36f19ea3126 +size 1867943218 diff --git a/previous_single_task/DualArmSendPingpong-v1/metadata.json b/previous_single_task/DualArmSendPingpong-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..9ec5671be7d6f212b4a76a528c56474ec06f86ab --- /dev/null +++ b/previous_single_task/DualArmSendPingpong-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmSendPingpong-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0040000.npz", + "checkpoint_sha256": "70924c6c7880fc34d61197d289cbbbbdc15a3e146339aed6b10af36f19ea3126", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 40000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "7538d14ae8f1548ebb177fa105ecca84b23a4eac6b45d8dcf463c9e16466d858", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmSendPingpong-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_dex/level_4/send_pingpong.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmSendPingpong-v1/norm_stats.json b/previous_single_task/DualArmSendPingpong-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..32bd83206dff9e14b78ddb88993afc7707abb4a0 --- /dev/null +++ b/previous_single_task/DualArmSendPingpong-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.007558815181255341, + -0.00462412228807807, + 0.018362829461693764, + 0.0, + 0.0, + 0.0, + -0.00023025987320579588, + 0.0019697367679327726, + 0.0009623697842471302, + 0.000459393544588238, + 0.016085918992757797, + 0.007613652851432562, + 0.0033041038550436497, + 0.00047639821423217654, + -0.0172930508852005, + -0.004068664740771055, + -0.0033155628480017185, + -0.00003764904249692336, + -0.018049336969852448, + 0.00018491584341973066, + 0.0010797119466587901, + -0.0005664853961206973, + 0.0071593597531318665, + 0.007843808270990849, + 0.003190523013472557, + 0.0031546640675514936, + 0.002360854996368289, + 0.0, + -0.0000551940975128673, + 0.024763869121670723, + 0.022655706852674484, + 0.022448597475886345, + 0.0001255798852071166, + 0.028918031603097916, + 0.03518712893128395, + 0.031011704355478287, + 0.00011533340148162097, + 0.02848169393837452, + 0.03339315205812454, + 0.029859693720936775, + 0.2674044966697693, + -0.00006691004819003865, + 0.15949541330337524, + 0.11673473566770554 + ], + "std": [ + 0.02168739214539528, + 0.01890762895345688, + 0.03509288281202316, + 0.0, + 0.0, + 0.0, + 0.009557455778121948, + 0.12242501974105835, + 0.009900037199258804, + 0.014271036721765995, + 0.011987438425421715, + 0.1206740289926529, + 0.005177566781640053, + 0.014383537694811821, + 0.03891371190547943, + 0.11488141119480133, + 0.085713230073452, + 0.046963632106781006, + 0.12529563903808594, + 0.018780872225761414, + 0.014820385724306107, + 0.011001449078321457, + 0.027551399543881416, + 0.032659564167261124, + 0.09630483388900757, + 0.016282977536320686, + 0.0115428501740098, + 0.0, + 0.00009782052075024694, + 0.08389901369810104, + 0.08383417874574661, + 0.08359035849571228, + 0.00022778402490075678, + 0.08285532146692276, + 0.08106328547000885, + 0.08160561323165894, + 0.0004331502423156053, + 0.08301010727882385, + 0.08142435550689697, + 0.08188889920711517, + 0.12564994394779205, + 0.0007427617092616856, + 0.08874961733818054, + 0.07931794226169586 + ], + "q01": [ + -0.05000000074505806, + -0.05000000074505806, + -0.05000000074505806, + 0.0, + 0.0, + 0.0, + -0.038537095163762566, + -0.501004667365551, + -0.038516724456846715, + -0.05780981673449278, + -5.329556188371498e-6, + -0.48920595504045483, + -0.014699050591140985, + -0.05813278920948506, + -0.15416394673585893, + -0.4736052021980286, + -0.35377475202083586, + -0.19270759944915772, + -0.6742515832185745, + -0.07702622083723545, + -0.07596274788975715, + -0.05757022456228733, + -0.05000000074505806, + -0.05020383580029011, + -0.07909566615223884, + -0.005823672517389057, + -0.011704011242091653, + 0.0, + -0.00044394970046560047, + -0.000020460169253055938, + -2.7418136596679688e-6, + -5.960464477539062e-7, + -0.0007975521026761271, + -0.000019918767065973952, + -2.4909609237511177e-6, + -1.1918801015586844e-11, + -0.0018437557809054855, + -0.00001937669912877027, + -2.4228941128967563e-6, + -1.726499672394244e-11, + -2.3625225367140956e-6, + -0.0006171230343170464, + -6.226533059816575e-6, + -6.438860111757094e-8 + ], + "q99": [ + 0.04998000074476004, + 0.04998000074476004, + 0.04998000074476004, + 0.0, + 0.0, + 0.0, + 0.03852569177150726, + 0.5002075980186462, + 0.03847609116435051, + 0.05778003284037114, + 0.029503009120873867, + 0.5002003032445907, + 0.019232291615754363, + 0.057760441121459005, + 0.13398102445304394, + 0.4617048907279968, + 0.34627731349468227, + 0.19259901466369628, + 0.2517912715673445, + 0.07704410968124867, + 0.025004938399791718, + 0.02249367671310902, + 0.04998000074476004, + 0.10651465903967619, + 0.6489090492546558, + 0.09673828838765622, + 0.06569377581477165, + 0.0, + 0.00004888547340960929, + 0.5007409111994333, + 0.5007576071739197, + 0.5007598447799683, + 0.0006744107229518705, + 0.5007415406189772, + 0.5007577106891262, + 0.5007599816294961, + 0.0007053828600794071, + 0.5007421242719589, + 0.5007578181561025, + 0.5007599816282686, + 0.3851180496287705, + 0.0021568325679283593, + 0.5003692824371162, + 0.5007599668486098 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmSendPingpong-v1/source_audit.json b/previous_single_task/DualArmSendPingpong-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..7bc3a139d04cee712eb64a6b223ab0583f390fe2 --- /dev/null +++ b/previous_single_task/DualArmSendPingpong-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmSendPingpong-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.15", + "peak_step": "35000", + "final": "0.0", + "start": "2026-08-22 06:21:04", + "eval_mtime": "2026-08-25 13:01:59.472132", + "source_mtime": "2026-08-02 15:19:53.067536", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_dex/level_4/send_pingpong.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmSendPingpong-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmSendPingpong-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmSendPingpong-v1/lora_params/lora_step_0035000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmSetupTable-v1/eval_result.json b/previous_single_task/DualArmSetupTable-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..b1242b07b58849d223380fce5a070c6eb214745c --- /dev/null +++ b/previous_single_task/DualArmSetupTable-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmSetupTable-v1", + "level": "level_3", + "suite": "rigid_body/dual_arm_dex", + "step": 25000, + "episodes": 50, + "successes": 2, + "success_rate": 0.04, + "return_mean": 963.8013223078847, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmSetupTable-v1/evaluation_history.csv b/previous_single_task/DualArmSetupTable-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..982534a8c2722a08c09ea7b29d9d00cee9cd9a9c --- /dev/null +++ b/previous_single_task/DualArmSetupTable-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,66.63242879420518,True +10000,50,0,0.0,664.2008750444651,True +15000,50,0,0.0,60.804537423849105,True +20000,50,0,0.0,226.96615758299828,True +25000,50,2,0.04,963.8013223078847,True +30000,50,0,0.0,130.08275908350944,True +35000,50,0,0.0,66.73280980020762,True +40000,50,2,0.04,1395.3220165485143,True diff --git a/previous_single_task/DualArmSetupTable-v1/lora_step_0025000.npz b/previous_single_task/DualArmSetupTable-v1/lora_step_0025000.npz new file mode 100644 index 0000000000000000000000000000000000000000..73969d12b93106d9ec7ccc6880847afd26d04cc2 --- /dev/null +++ b/previous_single_task/DualArmSetupTable-v1/lora_step_0025000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:093e094f8ea395f959f4160b04c57930e33c5a3df4cc6504428a877b03a2244c +size 1867992394 diff --git a/previous_single_task/DualArmSetupTable-v1/metadata.json b/previous_single_task/DualArmSetupTable-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..fb67bcbfdcfcd16bb8b2de6e35b6fe0df2f50285 --- /dev/null +++ b/previous_single_task/DualArmSetupTable-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmSetupTable-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0025000.npz", + "checkpoint_sha256": "093e094f8ea395f959f4160b04c57930e33c5a3df4cc6504428a877b03a2244c", + "checkpoint_bytes": 1867992394, + "tensor_count": 51, + "action_dim": 50, + "state_dim": 50, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 25000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "7e9ff7839b9613bd12dcf34b729b78461b741eaa4f1c6a09a6fc76b1c6073ec9", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmSetupTable-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_dex/level_3/setup_table.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmSetupTable-v1/norm_stats.json b/previous_single_task/DualArmSetupTable-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..47e3a16437fcd5a79c361b2c660ece4781e3baa8 --- /dev/null +++ b/previous_single_task/DualArmSetupTable-v1/norm_stats.json @@ -0,0 +1,424 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -1.2934833765029907, + 0.20288757979869843, + 0.00011599037679843605, + 0.02894383855164051, + -0.024871813133358955, + -0.0007358205621130764, + 0.002233327366411686, + 0.002123454352840781, + 0.006865745410323143, + 0.0008681290782988071, + 0.0007038091425783932, + 0.0012731124879792333, + 0.0005659931339323521, + 0.0005933192442171276, + 0.019272077828645706, + 0.011318647302687168, + 0.004750886466354132, + -0.0008551024366170168, + 0.00194376555737108, + 0.001273082452826202, + 0.0005659951129928231, + 0.08293803781270981, + -1.7359883486278704e-6, + 0.03385690227150917, + 0.00042426277650520205, + -0.2769774794578552, + -0.8796277046203613, + -0.0003567124658729881, + 0.020280208438634872, + -0.000631770002655685, + -0.0006392644136212766, + -0.008384406566619873, + 0.002487919759005308, + -0.006908698473125696, + -0.0008721950580365956, + 0.0027467242907732725, + 0.0012789927423000336, + 0.0005703548085875809, + 0.0029668661300092936, + 0.026822539046406746, + 0.02574099600315094, + 0.013943188823759556, + -0.006017501000314951, + 0.024296889081597328, + 0.027483170852065086, + 0.015599929727613926, + -0.05520467460155487, + -0.004279676359146833, + 0.11798181384801865, + 0.047938309609889984 + ], + "std": [ + 0.00562053918838501, + 0.24423116445541382, + 0.0003551544505171478, + 0.06621827930212021, + 0.06279098987579346, + 0.04510004073381424, + 0.029521141201257706, + 0.004433004185557365, + 0.017168283462524414, + 0.025708476081490517, + 0.02005532570183277, + 0.03612646088004112, + 0.01605718955397606, + 0.0032695478294044733, + 0.034700993448495865, + 0.043641041964292526, + 0.018648764118552208, + 0.02570156194269657, + 0.02136876806616783, + 0.03612656518816948, + 0.016057195141911507, + 0.1438012570142746, + 4.243704552209238e-6, + 0.154630646109581, + 0.012043356895446777, + 0.006039705593138933, + 0.4723264276981354, + 0.0008576271357014775, + 0.06438086926937103, + 0.07822326570749283, + 0.12946321070194244, + 0.04177717864513397, + 0.005956397391855717, + 0.014689992181956768, + 0.025706859305500984, + 0.020409289747476578, + 0.12234841287136078, + 0.11828313022851944, + 0.004722921643406153, + 0.034119557589292526, + 0.12519872188568115, + 0.1185184195637703, + 0.028920117765665054, + 0.032164402306079865, + 0.12201616168022156, + 0.11678416281938553, + 0.23637495934963226, + 0.02524641528725624, + 0.22893615067005157, + 0.14991864562034607 + ], + "q01": [ + -1.293515207004547, + -0.22494162547588348, + 0.0, + -0.00014839372038841247, + -0.30000001192092896, + -0.2588400102853775, + -0.011919230953231453, + -0.004671362254023552, + -0.020900992995500564, + -0.07881710221320391, + -0.06408695520162583, + -0.11540655872821808, + -0.05129369588494301, + -0.00657828969694674, + -0.00009341618083417384, + -0.09603913065493107, + -0.040777081601321694, + -0.09284374649375678, + -0.06308718710541725, + -0.11540700209140778, + -0.05129369567036629, + -0.09316579377651214, + -0.000011938640293374192, + -0.32886609926223753, + -0.038470238649845126, + -0.2814531555175781, + -1.4002350568771362, + -0.0028307230494916268, + -0.03452903479337692, + -0.30000001192092896, + -0.30000001192092896, + -0.30000001192092896, + -0.008417095512151718, + -0.05118781199678779, + -0.09284356041699648, + -0.06412934268712997, + -0.3361011236190796, + -0.3361004418373108, + -0.000021174597350182012, + -0.03972552351951599, + -0.2551107620060444, + -0.28347658851146695, + -0.08026614533513785, + -0.042808899316191674, + -0.24214828729629517, + -0.27325377615690233, + -0.4357578517377377, + -0.07376355617195368, + -0.33435247659683226, + -0.19373404664993288 + ], + "q99": [ + -1.293499950003624, + 0.6553590275228023, + 0.0012973557816352696, + 0.29993984988331795, + -0.000010861754417390568, + 0.19212000763416293, + 0.190977840314433, + 0.017315022632479665, + 0.06683734422773123, + 0.09268927109688518, + 0.06613833869695665, + 0.11920892065167427, + 0.053008686012029654, + 0.011380596198141576, + 0.09797912821248174, + 0.1386593349739909, + 0.056236493714153776, + 0.0786938293725252, + 0.0688929494202137, + 0.1192089904397726, + 0.05300868951678275, + 0.3578615966945886, + 8.100136404755177e-6, + 0.5195188292503357, + 0.0397215859681368, + -0.26350357303619387, + 0.037284882023930654, + -6.13773427681711e-7, + 0.29993103685975075, + 0.17412000691890717, + 0.29988001191616065, + 0.014866637811064731, + 0.02894875770807266, + 0.011189182385802271, + 0.07860646246671676, + 0.0663794688552618, + 0.5275618901371956, + 0.5275619225859642, + 0.018009167671438626, + 0.0986231538414955, + 0.5276623317599297, + 0.527717272233963, + 0.0906135828629136, + 0.07756568507552147, + 0.5277013698101044, + 0.5276852785408498, + 0.8553009045302868, + 0.10687841912806034, + 0.6082476333916188, + 0.5886311592340469 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmSetupTable-v1/source_audit.json b/previous_single_task/DualArmSetupTable-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..c17749f0d69f20830edafb9c05bdd83ed6134b45 --- /dev/null +++ b/previous_single_task/DualArmSetupTable-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmSetupTable-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.15", + "peak_step": "25000", + "final": "0.0", + "start": "2026-08-22 06:41:01", + "eval_mtime": "2026-08-25 12:04:47.210212", + "source_mtime": "2026-08-21 12:25:44.022796", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_dex/level_3/setup_table.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmSetupTable-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmSetupTable-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmSetupTable-v1/lora_params/lora_step_0025000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmShapeDoll-v1/eval_result.json b/previous_single_task/DualArmShapeDoll-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..a9519ab0273a488cdd6e0d15ef9e9ed22d4a978d --- /dev/null +++ b/previous_single_task/DualArmShapeDoll-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmShapeDoll-v1", + "level": "level_3", + "suite": "deformable/dual_arm_gripper", + "step": 30000, + "episodes": 50, + "successes": 2, + "success_rate": 0.04, + "return_mean": 438.03986117392776, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmShapeDoll-v1/evaluation_history.csv b/previous_single_task/DualArmShapeDoll-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..35c1a7936631a9d2559abf86671be3864c2c6a4f --- /dev/null +++ b/previous_single_task/DualArmShapeDoll-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,440.2142948436737,True +10000,50,0,0.0,433.5211352854967,True +15000,50,0,0.0,439.71059030920264,True +20000,50,0,0.0,448.3843006896973,True +25000,50,1,0.02,455.72908428907397,True +30000,50,2,0.04,438.03986117392776,True +35000,50,0,0.0,455.52062763601543,True +40000,50,0,0.0,453.87199902892115,True diff --git a/previous_single_task/DualArmShapeDoll-v1/lora_step_0030000.npz b/previous_single_task/DualArmShapeDoll-v1/lora_step_0030000.npz new file mode 100644 index 0000000000000000000000000000000000000000..f7f219be094275067c995c5402f4750678c5cf6d --- /dev/null +++ b/previous_single_task/DualArmShapeDoll-v1/lora_step_0030000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e290f251fb1feb05077deb5167416a0bf34c273229cff2ad60b0dfbd91934955 +size 1867844866 diff --git a/previous_single_task/DualArmShapeDoll-v1/metadata.json b/previous_single_task/DualArmShapeDoll-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..3fc8baf8be1f288eb8ebd2e4ec795cbcb2c0ef8a --- /dev/null +++ b/previous_single_task/DualArmShapeDoll-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmShapeDoll-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0030000.npz", + "checkpoint_sha256": "e290f251fb1feb05077deb5167416a0bf34c273229cff2ad60b0dfbd91934955", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 30000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "be52505b572ddfa20e1be5b3219582a4652ab49e824cae042981c394e5d264ef", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmShapeDoll-v1", + "source_path": "maniladder/envs/deformable/dual_arm_gripper/level_3/shape_doll.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmShapeDoll-v1/norm_stats.json b/previous_single_task/DualArmShapeDoll-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..b5a4800572402970e678b99a3d2891516a482b41 --- /dev/null +++ b/previous_single_task/DualArmShapeDoll-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.0035847220569849014, + -0.003022491466253996, + -0.0002696212031878531, + -6.341006155707873e-7, + 0.00015174482541624457, + 0.0008939015679061413, + 0.11106375604867935, + -0.0006169121479615569, + 0.00238724029622972, + -0.00011975658708252013, + -0.0002505740849301219, + 0.0005169024225324392, + 0.0004901227657683194, + 0.1815384477376938, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.043701451271772385, + 0.057592958211898804, + 0.02533373236656189, + 0.02490021474659443, + 0.025277895852923393, + 0.07468754053115845, + 0.9301884770393372, + 0.039843469858169556, + 0.05695484206080437, + 0.02573854848742485, + 0.016473345458507538, + 0.015349538065493107, + 0.07213912904262543, + 0.9079586267471313, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.11999999731779099, + -0.11999999731779099, + -0.07000000029802322, + -0.030800000131130212, + -0.010864000046253208, + -0.2800000011920929, + -1.0, + -0.11999999731779099, + -0.11999999731779099, + -0.07000000029802322, + -0.005145370018482198, + -0.0047931740701198655, + -0.2800000011920929, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.11995199731886386, + 0.11995199731886386, + 0.06997200029790401, + 0.029456000125408166, + 0.03796800016164781, + 0.27988800119161605, + 0.9996, + 0.11995199731886386, + 0.11995199731886386, + 0.06997200029790401, + 0.011826407527923577, + 0.022424335539340978, + 0.27988800119161605, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmShapeDoll-v1/source_audit.json b/previous_single_task/DualArmShapeDoll-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..fe9a3aedb95f814ccbda313ccfd028966067c537 --- /dev/null +++ b/previous_single_task/DualArmShapeDoll-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmShapeDoll-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-09-04 06:04:18", + "eval_mtime": "2026-09-07 12:48:51.150128", + "source_mtime": "2026-08-25 07:33:36.167211", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_gripper/level_3/shape_doll.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmShapeDoll-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmShapeDoll-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmShapeDoll-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmShootBasketball-v1/eval_result.json b/previous_single_task/DualArmShootBasketball-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..b0e591cb796772f999ef24b0b6c499106ed16935 --- /dev/null +++ b/previous_single_task/DualArmShootBasketball-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmShootBasketball-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_dex", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": -124.416729821607, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmShootBasketball-v1/evaluation_history.csv b/previous_single_task/DualArmShootBasketball-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..4524abf8b4a22273e28ed89390e9cef3f7417aa0 --- /dev/null +++ b/previous_single_task/DualArmShootBasketball-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,-124.416729821607,True +10000,50,0,0.0,-128.97785466102417,True +15000,50,0,0.0,-123.21594117723347,True +20000,50,0,0.0,-119.13311796374502,True +25000,50,0,0.0,-119.39706527374976,True +30000,50,0,0.0,-115.1265727031979,True +35000,50,0,0.0,-118.29808492508077,True +40000,50,0,0.0,-117.65816518728622,True diff --git a/previous_single_task/DualArmShootBasketball-v1/lora_step_0005000.npz b/previous_single_task/DualArmShootBasketball-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..ecb71e1b80045e225fc4c21568d247fb9c2b6800 --- /dev/null +++ b/previous_single_task/DualArmShootBasketball-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:03ce4d67db10529e152c37e5894df057c430f38592273d6ec47cbbd4500226e7 +size 1867992394 diff --git a/previous_single_task/DualArmShootBasketball-v1/metadata.json b/previous_single_task/DualArmShootBasketball-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..0ef110cffd4e93fc495b30b9ac87a16af26c2671 --- /dev/null +++ b/previous_single_task/DualArmShootBasketball-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmShootBasketball-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "03ce4d67db10529e152c37e5894df057c430f38592273d6ec47cbbd4500226e7", + "checkpoint_bytes": 1867992394, + "tensor_count": 51, + "action_dim": 50, + "state_dim": 50, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "4e0c526c93208b4f65b0f619c76bda31b2941ff5806f0acc243d42ade793067f", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmShootBasketball-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_dex/level_4/shoot_basketball.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmShootBasketball-v1/norm_stats.json b/previous_single_task/DualArmShootBasketball-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..8e872ce0e12f40c001090b2deca2603e082941d0 --- /dev/null +++ b/previous_single_task/DualArmShootBasketball-v1/norm_stats.json @@ -0,0 +1,424 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0991581529378891, + -1.2232592105865479, + 1.6868301630020142, + -0.00016725290333852172, + 0.00010371019743615761, + 0.003165945177897811, + 0.00006541863695019856, + 0.000023097732992027886, + 0.018175702542066574, + 0.002868842566385865, + 0.004283479414880276, + 0.00529110711067915, + 0.002337581478059292, + -0.0015768500743433833, + 0.014184821397066116, + 0.010287553071975708, + 0.0038991570472717285, + -0.0027387787122279406, + 0.008448445238173008, + 0.006605592556297779, + 0.002619293751195073, + 0.03286903351545334, + -0.010726574808359146, + 0.0050604804418981075, + 0.022139044478535652, + 0.03387391194701195, + 1.3330473899841309, + 1.6855380535125732, + -0.00006477859278675169, + -0.0005730738630518317, + 0.02515808306634426, + -0.0005534743540920317, + -0.00014024131814949214, + -0.025371599942445755, + -0.0011916738003492355, + 0.013024941086769104, + 0.00996445957571268, + 0.003980010747909546, + -0.00024222592764999717, + 0.017173249274492264, + 0.013043839484453201, + 0.0050893728621304035, + 0.012927649542689323, + 0.01866978034377098, + 0.009494463913142681, + 0.0038533667102456093, + 0.03135930746793747, + -0.007313848473131657, + -0.2255406677722931, + 0.0030081907752901316 + ], + "std": [ + 1.0916107892990112, + 0.38945916295051575, + 1.5566240549087524, + 0.08127929270267487, + 0.10277026146650314, + 0.06793590635061264, + 0.027579886838793755, + 0.016621701419353485, + 0.05994713306427002, + 0.017015354707837105, + 0.026884377002716064, + 0.03453754261136055, + 0.015356061980128288, + 0.0017749544931575656, + 0.029610516503453255, + 0.03762250766158104, + 0.014927723444998264, + 0.01700102724134922, + 0.027150452136993408, + 0.03431921452283859, + 0.015307723544538021, + 0.17262759804725647, + 0.059539344161748886, + 0.010151023976504803, + 0.1447831094264984, + 1.0371538400650024, + 0.6918918490409851, + 1.556729793548584, + 0.08848723769187927, + 0.06836457550525665, + 0.08782348036766052, + 0.07296161353588104, + 0.012161863036453724, + 0.05959458276629448, + 0.016464196145534515, + 0.021244358271360397, + 0.0210097823292017, + 0.00917579885572195, + 0.0019527454860508442, + 0.025089973583817482, + 0.024518219754099846, + 0.009626968763768673, + 0.07538163661956787, + 0.04173188656568527, + 0.03762707859277725, + 0.016540465876460075, + 0.1594405323266983, + 0.011307036504149437, + 0.11290306597948074, + 0.00439170328900218 + ], + "q01": [ + -1.685032844543457, + -1.8125729485988618, + -2.220446049250313e-16, + -0.0391999999999999, + -0.24, + -0.24008116381168365, + -0.05700000226497648, + -0.06813475585579871, + -0.21984000873565673, + -0.025152810126543045, + -0.0909546236038208, + -0.11734981667995453, + -0.052188566753268245, + -0.005886162966562552, + -0.0630451711267233, + -0.116345061352849, + -0.047441944144666195, + -0.0916771925650537, + -0.0797220981106162, + -0.11348386253118514, + -0.05150300868153572, + -0.34495360931158064, + -0.23680134204030037, + -0.000045999771828064695, + -0.4925336833834648, + -1.684999942779541, + -0.00003170967102050781, + -0.006204571109265089, + -0.24, + -0.24020675137042996, + -0.24021686187088492, + -0.30000001192092896, + -0.03264327511191367, + -0.23051795887947082, + -0.09166644318774342, + -0.0001205442063510418, + -0.006326642590016127, + -0.0067521732814610005, + -0.006223018184304237, + -0.00012238264181651176, + -0.0003503413705388083, + -0.004060277629736811, + -0.27187432861328126, + -0.007768300835043192, + -0.11724165329039096, + -0.052207765257358554, + -0.3536217524528503, + -0.0472875751839907, + -0.3143803749157116, + -0.000010404533937980886 + ], + "q99": [ + 1.6848841621398924, + -0.0003492185354234678, + 3.147630224132538, + 0.04039999999999999, + 0.23960000000000004, + 0.07990865084826948, + 0.06624000263214114, + 0.07098168113827705, + 0.25740001022815706, + 0.09162756711244582, + 0.10529426364302635, + 0.13536890320777895, + 0.060160081338882454, + 0.000010498271149117326, + 0.10596471389234063, + 0.15109893148243425, + 0.06040693140774965, + 0.025113372117280985, + 0.10588002511560918, + 0.13557148923873902, + 0.060187499171495434, + 0.5173947797715664, + 0.19658296057581903, + 0.03740803672805705, + 0.5653466832160949, + 1.6843259428024289, + 2.3226437339305877, + 3.140962943179254, + 0.23960000000000004, + 0.23982557150125505, + 0.23987114653885355, + 0.29988001191616065, + 0.026577724784612677, + 0.015180626273155196, + 0.015819459833204735, + 0.09161936704069376, + 0.11776472622994333, + 0.052355655806884165, + 0.00549391770362854, + 0.0916190428766422, + 0.13089221886470914, + 0.05236371937794611, + 0.2693500473022461, + 0.20555676260776817, + 0.13984213197529316, + 0.06281626201570034, + 0.6402445590972901, + 0.000011741214770877306, + -0.00002171606309714358, + 0.017908914031738825 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmShootBasketball-v1/source_audit.json b/previous_single_task/DualArmShootBasketball-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..f3a681af9c2aae3cd667f98d49759fad3d843265 --- /dev/null +++ b/previous_single_task/DualArmShootBasketball-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmShootBasketball-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 06:21:06", + "eval_mtime": "2026-08-25 10:02:19.551790", + "source_mtime": "2026-05-15 06:04:05.969150", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_dex/level_4/shoot_basketball.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmShootBasketball-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmShootBasketball-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmShootBasketball-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmStuffCloth-v1/eval_result.json b/previous_single_task/DualArmStuffCloth-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..b7c4ecc74403a1a518f050e94c24150d66d43810 --- /dev/null +++ b/previous_single_task/DualArmStuffCloth-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmStuffCloth-v1", + "level": "level_2", + "suite": "deformable/dual_arm_dex", + "step": 15000, + "episodes": 50, + "successes": 20, + "success_rate": 0.4, + "return_mean": 46.894351813681425, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmStuffCloth-v1/evaluation_history.csv b/previous_single_task/DualArmStuffCloth-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..c04a603a3c6cbcb4789bfb7e4baca40d01e5c124 --- /dev/null +++ b/previous_single_task/DualArmStuffCloth-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,13,0.26,36.1589380820794,True +10000,50,5,0.1,12.514477247553877,True +15000,50,20,0.4,46.894351813681425,True +20000,50,16,0.32,41.14128454748308,True +25000,50,11,0.22,30.477255163099617,True +30000,50,16,0.32,20.593299287557603,True +35000,50,7,0.14,20.886216926290654,True +40000,50,13,0.26,23.812841891734863,True diff --git a/previous_single_task/DualArmStuffCloth-v1/lora_step_0015000.npz b/previous_single_task/DualArmStuffCloth-v1/lora_step_0015000.npz new file mode 100644 index 0000000000000000000000000000000000000000..ffde28bce60a1b1e0ea5486ed235fb13222a4389 --- /dev/null +++ b/previous_single_task/DualArmStuffCloth-v1/lora_step_0015000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:727b90f7ce8ebdf643ce17e321bf0546117642cc589645f96c2c941e6bb1a7a9 +size 1867943218 diff --git a/previous_single_task/DualArmStuffCloth-v1/metadata.json b/previous_single_task/DualArmStuffCloth-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..3d7b8a5fa22f0cb16dbce7304579eb4c15bea6b5 --- /dev/null +++ b/previous_single_task/DualArmStuffCloth-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmStuffCloth-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0015000.npz", + "checkpoint_sha256": "727b90f7ce8ebdf643ce17e321bf0546117642cc589645f96c2c941e6bb1a7a9", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 15000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "1af4854d0fc58f61f262babcf1aa05b2e79a5933e77dc5d0c6223d87617bc82b", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmStuffCloth-v1", + "source_path": "maniladder/envs/deformable/dual_arm_dex/level_2/stuff_cloth.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmStuffCloth-v1/norm_stats.json b/previous_single_task/DualArmStuffCloth-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..45da70767739fa733ad43eadc38ad420e83084dd --- /dev/null +++ b/previous_single_task/DualArmStuffCloth-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.0152354147285223, + -0.019486263394355774, + -0.021876685321331024, + 0.0, + 0.0, + 0.0, + -0.00007989671576069668, + 0.0031441058963537216, + 0.0027680329512804747, + 0.002601499669253826, + -0.00022614392219111323, + 0.004346957430243492, + 0.0031435538548976183, + 0.002631475916132331, + 0.0006732857436873019, + 0.010066886432468891, + 0.0055352807976305485, + 0.00322286831215024, + 0.002662173705175519, + -2.638426167322905e-6, + 0.0037329455371946096, + 0.0037374941166490316, + -0.007900726050138474, + 0.023532835766673088, + -0.002212431514635682, + 0.0, + 0.0, + 0.0, + 0.000023926662834128365, + 0.0678645521402359, + 0.032853566110134125, + 0.010595964267849922, + -0.000022421401808969676, + 0.008076941594481468, + 0.0023050468880683184, + 0.0020637037232518196, + -4.717820502264658e-6, + 0.0027551345992833376, + 0.0024360213428735733, + 0.0023159924894571304, + 0.04566705599427223, + -0.060469355434179306, + 0.1614563763141632, + 0.06483413279056549 + ], + "std": [ + 0.02482541650533676, + 0.028416695073246956, + 0.024844877421855927, + 0.0, + 0.0, + 0.0, + 0.00027917599072679877, + 0.008448999375104904, + 0.008473067544400692, + 0.008519187569618225, + 0.0005719992914237082, + 0.008695287629961967, + 0.008418997749686241, + 0.008518352173268795, + 0.003375234315171838, + 0.01275540329515934, + 0.00875450111925602, + 0.00840022787451744, + 0.008492871187627316, + 3.17646572511876e-6, + 0.011896232143044472, + 0.01189618743956089, + 0.04991554468870163, + 0.05584275349974632, + 0.07552523165941238, + 0.0, + 0.0, + 0.0, + 0.004325516056269407, + 0.12897434830665588, + 0.11972726881504059, + 0.12262239307165146, + 0.0000754117572796531, + 0.12354090064764023, + 0.12603574991226196, + 0.1258278489112854, + 0.00005721023626392707, + 0.12628856301307678, + 0.125972718000412, + 0.1258396953344345, + 0.1136188879609108, + 0.053912438452243805, + 0.21318486332893372, + 0.178549662232399 + ], + "q01": [ + -0.054999999701976776, + -0.054999999701976776, + -0.054999999701976776, + 0.0, + 0.0, + 0.0, + -0.0016630284816026688, + -0.0000576157197356224, + -0.000014513731002807617, + -0.0007615349203348161, + -0.0024981581484898926, + -0.00001052644699811936, + -2.696247398853305e-6, + -0.001571462407708168, + -0.0050007639806717635, + 0.00002110246568918228, + 1.385807991027832e-6, + -0.0013116618692874907, + 4.3213367462158203e-7, + -9.342156507108258e-6, + -0.000011444091796875, + 3.427267074584961e-6, + -0.17000000178813934, + -0.023267616415023805, + -0.08500000089406967, + 0.0, + 0.0, + 0.0, + -0.01209994544684887, + -0.3718808744430542, + -0.40033768928050995, + -0.43457207262516023, + -0.0003222069191688206, + -0.44790388016700744, + -0.47250254650115964, + -0.47250037231445313, + -0.00023809143668040633, + -0.4724040605068207, + -0.472607126480341, + -0.4725973863005638, + -0.29515075137615204, + -0.14047389977276326, + -0.5204325943350792, + -0.5962399766921997 + ], + "q99": [ + 0.019193091931939132, + 0.0044017396569252035, + -0.0001982181727886237, + 0.0, + 0.0, + 0.0, + 0.000012459386559203305, + 0.046014960773289204, + 0.04603216907680035, + 0.04597578680962324, + 0.0004916941644623873, + 0.04600757347345352, + 0.04602616202682257, + 0.046005674001574515, + 0.013345236573368311, + 0.04602059591561555, + 0.04602074874490499, + 0.04604231195598841, + 0.046020233951509, + 6.830397256863297e-9, + 0.0644498250424862, + 0.06445784962773322, + 0.16993200178742407, + 0.1699598381102085, + 0.16993200178742407, + 0.0, + 0.0, + 0.0, + 0.010846919715777042, + 0.42818521428108214, + 0.4282857662677765, + 0.42815803508758543, + 0.0000822189766331576, + 0.42810832524299613, + 0.42816334059238437, + 0.42816758613586436, + 0.0002086066422052682, + 0.42814226888418194, + 0.42817150501012813, + 0.4281761161088944, + 0.5498202690958977, + 0.005628071111440647, + 0.5497771444201469, + 0.5497600119233133 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmStuffCloth-v1/source_audit.json b/previous_single_task/DualArmStuffCloth-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..4c30310637c048bbfb8375c3c5e91b153c23576f --- /dev/null +++ b/previous_single_task/DualArmStuffCloth-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmStuffCloth-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.4", + "peak_step": "15000", + "final": "0.15", + "start": "2026-09-04 06:14:24", + "eval_mtime": "2026-09-08 00:44:12.460389", + "source_mtime": "2026-08-16 08:12:46.107465", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_dex/level_2/stuff_cloth.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmStuffCloth-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmStuffCloth-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmStuffCloth-v1/lora_params/lora_step_0015000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmThreadRopeThroughTube-v1/eval_result.json b/previous_single_task/DualArmThreadRopeThroughTube-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..ccd9ad849f9ee4172cdb558c1e56e819d5dff58e --- /dev/null +++ b/previous_single_task/DualArmThreadRopeThroughTube-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmThreadRopeThroughTube-v1", + "level": "level_3", + "suite": "deformable/dual_arm_gripper", + "step": 35000, + "episodes": 50, + "successes": 8, + "success_rate": 0.16, + "return_mean": 7.307999981641769, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmThreadRopeThroughTube-v1/evaluation_history.csv b/previous_single_task/DualArmThreadRopeThroughTube-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..523e78bd900bf4e5b61cdaf470ebeb09e5ebcae8 --- /dev/null +++ b/previous_single_task/DualArmThreadRopeThroughTube-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,5,0.1,16.00057137697935,True +10000,50,3,0.06,11.549714230298996,True +15000,50,4,0.08,9.501999981701374,True +20000,50,6,0.12,6.4322856843471525,True +25000,50,5,0.1,7.342571413218975,True +30000,50,5,0.1,10.025714241862298,True +35000,50,8,0.16,7.307999981641769,True +40000,50,7,0.14,13.937999838888645,True diff --git a/previous_single_task/DualArmThreadRopeThroughTube-v1/lora_step_0035000.npz b/previous_single_task/DualArmThreadRopeThroughTube-v1/lora_step_0035000.npz new file mode 100644 index 0000000000000000000000000000000000000000..5176aff0ab694a9b106f95f8bc3e6a71f6dba635 --- /dev/null +++ b/previous_single_task/DualArmThreadRopeThroughTube-v1/lora_step_0035000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ed640c43be58457fab4969a591ef4ab14103d5553c3f363f39412e0ff47bbdd1 +size 1867844866 diff --git a/previous_single_task/DualArmThreadRopeThroughTube-v1/metadata.json b/previous_single_task/DualArmThreadRopeThroughTube-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..7399eef9a4131e047465201c89c75fa2d319cff9 --- /dev/null +++ b/previous_single_task/DualArmThreadRopeThroughTube-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmThreadRopeThroughTube-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0035000.npz", + "checkpoint_sha256": "ed640c43be58457fab4969a591ef4ab14103d5553c3f363f39412e0ff47bbdd1", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 35000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "62084fd445a5402a2980d05e2dd0da77f8807d8c223a926a8e33c6a3a7f87394", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmThreadRopeThroughTube-v1", + "source_path": "maniladder/envs/deformable/dual_arm_gripper/level_3/through_tube.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmThreadRopeThroughTube-v1/norm_stats.json b/previous_single_task/DualArmThreadRopeThroughTube-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..0468b9664df53ff0539ea3ae34442dc4f8bbd600 --- /dev/null +++ b/previous_single_task/DualArmThreadRopeThroughTube-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.05168086290359497, + -0.05317944660782814, + -0.01167911384254694, + 0.0, + 0.0, + 0.011299656704068184, + -0.2725445628166199, + 0.0023345169611275196, + 0.03591533377766609, + -0.00657899584621191, + 0.0, + 0.0, + -0.001166406786069274, + 0.06584816426038742, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.07396484166383743, + 0.0838494747877121, + 0.09967145323753357, + 0.0, + 0.0, + 0.06920850276947021, + 0.9621431231498718, + 0.10051421821117401, + 0.1009552851319313, + 0.10139062255620956, + 0.0, + 0.0, + 0.05877416208386421, + 0.9978296160697937, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.18000000715255737, + -0.18000000715255737, + -0.18000000715255737, + 0.0, + 0.0, + -0.30155999486446383, + -1.0, + -0.18000000715255737, + -0.2199999988079071, + -0.18000000715255737, + 0.0, + 0.0, + -0.3499999940395355, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.1472400058507919, + 0.1399680055618286, + 0.1799280071496963, + 0.0, + 0.0, + 0.3498599940419197, + 0.9996, + 0.21991999880671503, + 0.17995999902486803, + 0.1799280071496963, + 0.0, + 0.0, + 0.16449999719858166, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmThreadRopeThroughTube-v1/source_audit.json b/previous_single_task/DualArmThreadRopeThroughTube-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..369c9a4c1ae5e2eac1dd88816a442b232274aa04 --- /dev/null +++ b/previous_single_task/DualArmThreadRopeThroughTube-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmThreadRopeThroughTube-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.15", + "peak_step": "5000", + "final": "0.1", + "start": "2026-09-04 06:04:20", + "eval_mtime": "2026-09-07 12:25:13.735260", + "source_mtime": "2026-08-23 08:48:11.192508", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_gripper/level_3/through_tube.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmThreadRopeThroughTube-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmThreadRopeThroughTube-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmThreadRopeThroughTube-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmThrowBlock-v1/eval_result.json b/previous_single_task/DualArmThrowBlock-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..6cbc764ebdb0a9a522e87d0d7a5c17d98a9d90e4 --- /dev/null +++ b/previous_single_task/DualArmThrowBlock-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmThrowBlock-v1", + "level": "level_2", + "suite": "rigid_body/dual_arm_dex", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": -806117.6206517029, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmThrowBlock-v1/evaluation_history.csv b/previous_single_task/DualArmThrowBlock-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..15769d1049217ff7dbcac6e9f09c7e0cad5e23c9 --- /dev/null +++ b/previous_single_task/DualArmThrowBlock-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,-806117.6206517029,True +10000,50,0,0.0,-22608.926204185485,True +15000,50,0,0.0,-15077.627834014893,True +20000,50,0,0.0,-25463.92994615555,True +25000,50,0,0.0,-5637.922432055473,True +30000,50,0,0.0,-36567.06400863647,True +35000,50,0,0.0,-26169.837536344527,True +40000,50,0,0.0,-7810.690715389252,True diff --git a/previous_single_task/DualArmThrowBlock-v1/lora_step_0005000.npz b/previous_single_task/DualArmThrowBlock-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..3e8cd03a730aacb47740d3b41e700aced6e8803e --- /dev/null +++ b/previous_single_task/DualArmThrowBlock-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:ddc085746228651a8ef69d51c548e10e6c50ca76081808df23890913a3bf20e0 +size 1867943218 diff --git a/previous_single_task/DualArmThrowBlock-v1/metadata.json b/previous_single_task/DualArmThrowBlock-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..c741a2dbade10677e9373d10fa3a1eae063e78ce --- /dev/null +++ b/previous_single_task/DualArmThrowBlock-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmThrowBlock-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "ddc085746228651a8ef69d51c548e10e6c50ca76081808df23890913a3bf20e0", + "checkpoint_bytes": 1867943218, + "tensor_count": 51, + "action_dim": 44, + "state_dim": 44, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "dc109157ea8ac22611196d93c02ae4d098d235f0e32e1bfb8a89cb8a616e31de", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmThrowBlock-v1", + "source_path": "maniladder/envs/rigid_body/dual_arm_dex/level_2/throw_ball.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmThrowBlock-v1/norm_stats.json b/previous_single_task/DualArmThrowBlock-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..c0a82c7218c80409821c165180588534212c5ffd --- /dev/null +++ b/previous_single_task/DualArmThrowBlock-v1/norm_stats.json @@ -0,0 +1,376 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.006455953698605299, + -0.018848862498998642, + -0.007671657484024763, + -0.007573489565402269, + 0.0, + 0.0, + 0.00047099695075303316, + 0.010609813034534454, + 0.009029376320540905, + 0.004022358451038599, + 1.3275498531584162e-6, + 0.00019726075697690248, + 0.0002188709913752973, + 0.00021222385112196207, + 1.0500092457732535e-6, + 0.00019823253387585282, + 0.00022072100546211004, + 0.0002141054137609899, + 0.013797201216220856, + 0.012767103500664234, + 0.021826285868883133, + 0.013724042102694511, + 0.0039455471560359, + 0.01673056371510029, + -0.008980144746601582, + 0.022720487788319588, + 0.0, + 0.0, + -1.2512292641986278e-6, + 0.012193078175187111, + 0.011300237849354744, + 0.011300494894385338, + -8.52245022997522e-7, + 0.011331302113831043, + 0.011329319328069687, + 0.011322179809212685, + -5.305394665811036e-7, + 0.011328368447721004, + 0.01132705993950367, + 0.011321122758090496, + 0.11780135333538055, + 0.00016152764146681875, + 0.06724559515714645, + 0.04977801814675331 + ], + "std": [ + 0.031874410808086395, + 0.036384545266628265, + 0.05512361600995064, + 0.10880747437477112, + 0.0, + 0.0, + 0.0014112000353634357, + 0.06771926581859589, + 0.0655083954334259, + 0.06484806537628174, + 3.941032446164172e-6, + 0.06608252972364426, + 0.06607811897993088, + 0.06607987731695175, + 4.2000510802608915e-6, + 0.06608280539512634, + 0.06607693433761597, + 0.06607834249734879, + 0.0689842477440834, + 0.029964584857225418, + 0.07810989022254944, + 0.06879021972417831, + 0.036202166229486465, + 0.036058034747838974, + 0.06908244639635086, + 0.18754471838474274, + 0.0, + 0.0, + 4.756111138704e-6, + 0.0993577241897583, + 0.09946976602077484, + 0.09947191923856735, + 4.70071836389252e-6, + 0.09945491701364517, + 0.09946514666080475, + 0.09946829080581665, + 5.556303221965209e-6, + 0.09945503622293472, + 0.09946537017822266, + 0.09946845471858978, + 0.13559949398040771, + 0.00044069989235140383, + 0.1090848445892334, + 0.10161078721284866 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + -0.0003141592741013888, + 0.0, + 0.0, + -0.00222569267777726, + -0.24091741671562195, + -0.2393708937168121, + -0.2610661326646805, + -8.09931536023214e-6, + -0.2786839356422424, + -0.27856289720535277, + -0.27857725906372066, + -9.44439324666746e-6, + -0.27848502527475355, + -0.2785473971366882, + -0.27855730056762695, + -0.22396528210639954, + -0.000014976741113059688, + -0.19929559235572816, + -0.22363914042711258, + -0.10000000149011612, + -0.007050336229801178, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -0.000017467455987934956, + -0.3563997752308845, + -0.35774586606025693, + -0.3577438385486603, + -0.000017455506179430813, + -0.35767598199844364, + -0.357696496963501, + -0.35770620942115783, + -0.000017446629305777606, + -0.35767870998382567, + -0.3576956734657287, + -0.35770570635795595, + -0.09665548291802406, + -0.00034987038630060854, + -0.2553939479589462, + -0.2873445487618446 + ], + "q99": [ + 0.09996000148952008, + -0.000014477728307246696, + 0.09996000148952008, + -0.0003141592741013888, + 0.0, + 0.0, + 0.005901063733454795, + 0.27838898994922634, + 0.2784793522357941, + 0.27850996798276895, + 0.00001127471314612194, + 0.2783211766958237, + 0.2783523164510727, + 0.2783471807003022, + 0.000011062878038501366, + 0.2783190510034561, + 0.2783570899724961, + 0.27835322022438047, + 0.2805299270629883, + 0.09192527266499037, + 0.2783327879428863, + 0.2783688572049141, + 0.09996000148952008, + 0.09997747406214476, + 0.09996000148952008, + 1.5704822112321852, + 0.0, + 0.0, + 6.84410737158032e-6, + 0.42242259991168973, + 0.4224506134629249, + 0.4224497791886329, + 7.237479005198111e-6, + 0.422466168308258, + 0.4224562061786651, + 0.4224539611935616, + 0.000014500061911530793, + 0.4224666598320007, + 0.4224564036607743, + 0.42245401710271835, + 0.32498961476087573, + 0.0014056117818690838, + 0.4224951197743416, + 0.4224810657501221 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmThrowBlock-v1/source_audit.json b/previous_single_task/DualArmThrowBlock-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..386ddeade42c51126251ea7a080d9b77108ef329 --- /dev/null +++ b/previous_single_task/DualArmThrowBlock-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmThrowBlock-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 06:31:04", + "eval_mtime": "2026-08-25 13:49:07.264684", + "source_mtime": "2026-04-30 05:45:56.913779", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/dual_arm_dex/level_2/throw_ball.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmThrowBlock-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmThrowBlock-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmThrowBlock-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/DualArmWeaveCloth-v1/eval_result.json b/previous_single_task/DualArmWeaveCloth-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..0f4f226306f8a05583aa79cd9533d7a816569949 --- /dev/null +++ b/previous_single_task/DualArmWeaveCloth-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "DualArmWeaveCloth-v1", + "level": "level_3", + "suite": "deformable/dual_arm_gripper", + "step": 20000, + "episodes": 50, + "successes": 16, + "success_rate": 0.32, + "return_mean": 111.53422757029533, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/DualArmWeaveCloth-v1/evaluation_history.csv b/previous_single_task/DualArmWeaveCloth-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..08942f36de25cf7441b9d9a24999fb2af7e88b12 --- /dev/null +++ b/previous_single_task/DualArmWeaveCloth-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,11,0.22,98.84533792033791,True +10000,50,6,0.12,90.26489328891039,True +15000,50,5,0.1,92.39555996343493,True +20000,50,16,0.32,111.53422757029533,True +25000,50,5,0.1,109.8071164008975,True +30000,50,10,0.2,102.55822728320956,True +35000,50,16,0.32,113.21956081196666,True +40000,50,9,0.18,117.1102277545631,True diff --git a/previous_single_task/DualArmWeaveCloth-v1/lora_step_0020000.npz b/previous_single_task/DualArmWeaveCloth-v1/lora_step_0020000.npz new file mode 100644 index 0000000000000000000000000000000000000000..85b5a4d9f3f3eff3fff97a452e16c462f491f85f --- /dev/null +++ b/previous_single_task/DualArmWeaveCloth-v1/lora_step_0020000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:0db96e943ccf5a38c41f132f289033e36e87358deb637445b03c9fdc37a1bd60 +size 1867844866 diff --git a/previous_single_task/DualArmWeaveCloth-v1/metadata.json b/previous_single_task/DualArmWeaveCloth-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..ef1038a3d4dfc607b5f4495890296eacfb8cd829 --- /dev/null +++ b/previous_single_task/DualArmWeaveCloth-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "DualArmWeaveCloth-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0020000.npz", + "checkpoint_sha256": "0db96e943ccf5a38c41f132f289033e36e87358deb637445b03c9fdc37a1bd60", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 20000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "256959eed6768e4e3a9747489cb6f60b0e1d6a350dcf79625257aaf48e2edac9", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_DualArmWeaveCloth-v1", + "source_path": "maniladder/envs/deformable/dual_arm_gripper/level_3/cloth_weave.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/DualArmWeaveCloth-v1/norm_stats.json b/previous_single_task/DualArmWeaveCloth-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..6e734112f538b62b830544cfb61fbc978c561b68 --- /dev/null +++ b/previous_single_task/DualArmWeaveCloth-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0007603890262544155, + -0.006732888985425234, + -0.0087584825232625, + 0.0, + 0.0, + 0.0, + -0.5583743453025818, + -0.0000544056310900487, + 0.010721794329583645, + -0.009998071007430553, + 0.0, + 0.0, + 0.0, + -0.5583743453025818, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.054586391896009445, + 0.048387035727500916, + 0.06382311135530472, + 0.0, + 0.0, + 0.0, + 0.8295891284942627, + 0.04677484557032585, + 0.06109045073390007, + 0.06268899142742157, + 0.0, + 0.0, + 0.0, + 0.8295891284942627, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.23999999463558197, + -0.23999999463558197, + -0.23999999463558197, + 0.0, + 0.0, + 0.0, + -1.0, + -0.23999999463558197, + -0.10999999940395355, + -0.23999999463558197, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.13992400059700016, + 0.05494099970310928, + 0.13992400059700016, + 0.0, + 0.0, + 0.0, + 0.9996, + 0.05494099970310928, + 0.23992999463677406, + 0.13992400059700016, + 0.0, + 0.0, + 0.0, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/DualArmWeaveCloth-v1/source_audit.json b/previous_single_task/DualArmWeaveCloth-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..e3710c44a13f23331410196409be25460454df95 --- /dev/null +++ b/previous_single_task/DualArmWeaveCloth-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "DualArmWeaveCloth-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.35", + "peak_step": "20000", + "final": "0.25", + "start": "2026-09-04 06:14:27", + "eval_mtime": "2026-09-07 12:32:56.986830", + "source_mtime": "2026-08-27 06:10:19.156978", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/dual_arm_gripper/level_3/cloth_weave.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmWeaveCloth-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_DualArmWeaveCloth-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_DualArmWeaveCloth-v1/lora_params/lora_step_0020000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/FitnessBall-v1/eval_result.json b/previous_single_task/FitnessBall-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..fadba15078ad1c6ff93c4137d00b9240e56a2ae5 --- /dev/null +++ b/previous_single_task/FitnessBall-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "FitnessBall-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_dex", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": -6493.543452997208, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/FitnessBall-v1/evaluation_history.csv b/previous_single_task/FitnessBall-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..c6dbca658404294a7cb53becc4d007fe71031fe3 --- /dev/null +++ b/previous_single_task/FitnessBall-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,-6493.543452997208,True +10000,50,0,0.0,-6459.505335321426,True +15000,50,0,0.0,-6479.7505559539795,True +20000,50,0,0.0,-6467.865859689712,True +25000,50,0,0.0,-6462.608285160065,True +30000,50,0,0.0,-6480.244885673523,True +35000,50,0,0.0,-6472.69580080986,True +40000,50,0,0.0,-6466.33156788826,True diff --git a/previous_single_task/FitnessBall-v1/lora_step_0005000.npz b/previous_single_task/FitnessBall-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..484ac9e4ae870d36f48005164004f55224d8e292 --- /dev/null +++ b/previous_single_task/FitnessBall-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:063779d861fa3fb188065506d615184f074752145694999e4bbf1f29220b18d9 +size 1867844866 diff --git a/previous_single_task/FitnessBall-v1/metadata.json b/previous_single_task/FitnessBall-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..bb3a2eed7d57fc72074c4279f031521611eb0b7a --- /dev/null +++ b/previous_single_task/FitnessBall-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "FitnessBall-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "063779d861fa3fb188065506d615184f074752145694999e4bbf1f29220b18d9", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "75bbcf587f1ecaae841f1355994d03b631efd25619a6e011ce2a66b11b358265", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_FitnessBall-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_4/fitness_ball.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/FitnessBall-v1/norm_stats.json b/previous_single_task/FitnessBall-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..b18dfbe4ab55c7301215d9296dac088909ab2919 --- /dev/null +++ b/previous_single_task/FitnessBall-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.009670303203165531, + 0.010119318962097168, + -0.005137340631335974, + -0.006285843439400196, + -0.010290244594216347, + -0.011907107196748257, + -0.002191759878769517, + 0.0003341968695167452, + 0.0015639543998986483, + 2.4573782297920843e-7, + 2.338688176450887e-7, + 1.70468285887182e-6, + 0.007295676041394472, + -0.01884879358112812, + 0.0227881520986557, + 0.03341180831193924, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.13150955736637115, + 0.016094941645860672, + 0.004109491128474474, + 0.00819680467247963, + 0.011581087484955788, + 0.11213427037000656, + 0.03722941130399704, + 0.09732692688703537, + 0.20365169644355774, + 4.6621786964351486e-7, + 1.0338985703128856e-6, + 3.3569787660781003e-7, + 0.05311492085456848, + 0.062381889671087265, + 0.09930748492479324, + 0.05628596991300583, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + -0.32524628192186356, + -0.03032128846645355, + -0.013867685985565186, + -0.02859351336956024, + -0.038541786920315194, + -0.26963016386032107, + -0.10097612795829773, + -0.22648115117549897, + -0.3995824721455574, + -5.960464477539062e-7, + -5.960464477539062e-7, + 1.667642593383789e-6, + -0.12412295017242432, + -0.16479217433929444, + -0.13269053936004638, + -0.12860755457878112, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.27888114750385284, + 0.044565224432945263, + -0.00015148131847381695, + -1.3363122940042438e-6, + -0.00017019071861842233, + 0.2520409151554108, + 0.08976618115901946, + 0.2684709525585175, + 0.46203023874759674, + 8.341789245605467e-7, + 4.648208618164062e-6, + 2.9794692993164063e-6, + 0.0950482222020626, + 0.07587548986077311, + 0.2863426017761231, + 0.17453809108734125, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/FitnessBall-v1/source_audit.json b/previous_single_task/FitnessBall-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..d61715ade37d11fbb703d4f1f2f79e9c4e952464 --- /dev/null +++ b/previous_single_task/FitnessBall-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "FitnessBall-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 06:21:08", + "eval_mtime": "2026-08-25 11:17:49.477225", + "source_mtime": "2026-08-02 15:03:41.487108", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_4/fitness_ball.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_FitnessBall-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_FitnessBall-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_FitnessBall-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/FlipPan-v1/eval_result.json b/previous_single_task/FlipPan-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..690424e11dffadc8498cc42aa7fd333f3efed6b6 --- /dev/null +++ b/previous_single_task/FlipPan-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "FlipPan-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_gripper", + "step": 40000, + "episodes": 50, + "successes": 2, + "success_rate": 0.04, + "return_mean": 1115.2483033573628, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/FlipPan-v1/evaluation_history.csv b/previous_single_task/FlipPan-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..dfc0232638b38b77cb3e3e377b25208bfd35b426 --- /dev/null +++ b/previous_single_task/FlipPan-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,1850.3694087743759,True +10000,50,0,0.0,1619.979951660037,True +15000,50,0,0.0,1598.9930662673712,True +20000,50,0,0.0,768.4725514292717,True +25000,50,0,0.0,1004.3939337867498,True +30000,50,0,0.0,1547.7505061429738,True +35000,50,0,0.0,1257.2300715255737,True +40000,50,2,0.04,1115.2483033573628,True diff --git a/previous_single_task/FlipPan-v1/lora_step_0040000.npz b/previous_single_task/FlipPan-v1/lora_step_0040000.npz new file mode 100644 index 0000000000000000000000000000000000000000..503f3418f2e11c85d4df387f1ac53fd8e5840a09 --- /dev/null +++ b/previous_single_task/FlipPan-v1/lora_step_0040000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4ef6168e12dd59bc8bd8d9dacf8e10df6be69cb341dc4d3c85b2b8c08cc0fa95 +size 1867844866 diff --git a/previous_single_task/FlipPan-v1/metadata.json b/previous_single_task/FlipPan-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..e486b2a60746eb66e860161a20df28a7a158cb68 --- /dev/null +++ b/previous_single_task/FlipPan-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "FlipPan-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0040000.npz", + "checkpoint_sha256": "4ef6168e12dd59bc8bd8d9dacf8e10df6be69cb341dc4d3c85b2b8c08cc0fa95", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 40000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "352cac949d76f0e5bdaa0e736494bbc6c6abc8af725b44290f01ff3a0443dd75", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_FlipPan-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_4/flip_pan.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/FlipPan-v1/norm_stats.json b/previous_single_task/FlipPan-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..ed8b2da87202c6c3c0bb73f45c7b85429f847362 --- /dev/null +++ b/previous_single_task/FlipPan-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.00370412296615541, + -0.0018775112694129348, + 0.015247388742864132, + -0.0008927933522500098, + -0.0003451751545071602, + -0.0018576805014163256, + -0.2139403223991394, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.060107458382844925, + 0.08555598556995392, + 0.5836642384529114, + 0.023278720676898956, + 0.11470771580934525, + 0.03747909888625145, + 0.9768467545509338, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.1566876028060913, + -0.2732511058807373, + -2.0, + -0.023368743038177542, + -0.3495455331087112, + -0.12030481081008909, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.22404728612899782, + 0.26105050048828127, + 1.9992, + 0.011924826765060348, + 0.34889602707624434, + 0.11980658693313595, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/FlipPan-v1/source_audit.json b/previous_single_task/FlipPan-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..136cda5ccedf4c75c466271258a02a972fef137a --- /dev/null +++ b/previous_single_task/FlipPan-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "FlipPan-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.05", + "peak_step": "40000", + "final": "0.05", + "start": "2026-08-21 13:40:34", + "eval_mtime": "2026-08-24 21:10:13.414941", + "source_mtime": "2026-08-07 08:33:13.993809", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_4/flip_pan.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_FlipPan-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_FlipPan-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_FlipPan-v1/lora_params/lora_step_0040000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/FoldShirt-v1/eval_result.json b/previous_single_task/FoldShirt-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..5541806dd1a48b15cc1cc6308b753b872f386081 --- /dev/null +++ b/previous_single_task/FoldShirt-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "FoldShirt-v1", + "level": "level_2", + "suite": "deformable/single_arm_gripper", + "step": 40000, + "episodes": 50, + "successes": 36, + "success_rate": 0.72, + "return_mean": 170.07801467537817, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/FoldShirt-v1/evaluation_history.csv b/previous_single_task/FoldShirt-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..c3f4fec29535f076e714a63e6f09624f42000b6b --- /dev/null +++ b/previous_single_task/FoldShirt-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,25,0.5,131.64590929369334,True +10000,50,25,0.5,77.56784317751968,True +15000,50,30,0.6,152.34612643880502,True +20000,50,31,0.62,140.93384002179926,True +25000,50,28,0.56,160.46879280454684,True +30000,50,25,0.5,111.71872924421727,True +35000,50,29,0.58,199.52652104569995,True +40000,50,36,0.72,170.07801467537817,True diff --git a/previous_single_task/FoldShirt-v1/lora_step_0040000.npz b/previous_single_task/FoldShirt-v1/lora_step_0040000.npz new file mode 100644 index 0000000000000000000000000000000000000000..bac5cd07d48deb34451d8755981f6a549bf70705 --- /dev/null +++ b/previous_single_task/FoldShirt-v1/lora_step_0040000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:594d272a50755c96e1a9cd31a8a961962a4b34c7295a64fc39ed716a4e19cbe0 +size 1867844866 diff --git a/previous_single_task/FoldShirt-v1/metadata.json b/previous_single_task/FoldShirt-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..c513fa596cd369cf49750545e48c4ac040581256 --- /dev/null +++ b/previous_single_task/FoldShirt-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "FoldShirt-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0040000.npz", + "checkpoint_sha256": "594d272a50755c96e1a9cd31a8a961962a4b34c7295a64fc39ed716a4e19cbe0", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 40000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "173ec60032a5fc7904ff7ba581b2a08c80263a88ebff59b89ea5340c7541668b", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_FoldShirt-v1", + "source_path": "maniladder/envs/deformable/single_arm_gripper/level_2/fold_shirt.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/FoldShirt-v1/norm_stats.json b/previous_single_task/FoldShirt-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..2e628afc9fd2195e706c0e371d01d401a174d3cc --- /dev/null +++ b/previous_single_task/FoldShirt-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.006568667944520712, + -0.014366468414664268, + -0.022168491035699844, + 0.0, + 0.0, + 0.0, + 0.29754698276519775, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.03817269951105118, + 0.05037758871912956, + 0.07240109145641327, + 0.0, + 0.0, + 0.0, + 0.9527692794799805, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10003199776411056, + -0.11999999731779099, + -0.11999999731779099, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.11995199731886386, + 0.09998399776518344, + 0.11995199731886386, + 0.0, + 0.0, + 0.0, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/FoldShirt-v1/source_audit.json b/previous_single_task/FoldShirt-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..4f69af088de6491a4094620cf677283fbdc757a5 --- /dev/null +++ b/previous_single_task/FoldShirt-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "FoldShirt-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.85", + "peak_step": "15000", + "final": "0.65", + "start": "2026-09-04 06:04:22", + "eval_mtime": "2026-09-07 12:30:41.080084", + "source_mtime": "2026-08-16 05:16:16.102128", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/single_arm_gripper/level_2/fold_shirt.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_FoldShirt-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_FoldShirt-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_FoldShirt-v1/lora_params/lora_step_0015000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/GraspCylinder-v1/eval_result.json b/previous_single_task/GraspCylinder-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..b3362ab70933694e8e2095c15182961d3a6511af --- /dev/null +++ b/previous_single_task/GraspCylinder-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "GraspCylinder-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_dex", + "step": 15000, + "episodes": 50, + "successes": 47, + "success_rate": 0.94, + "return_mean": 25.653659981940546, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/GraspCylinder-v1/evaluation_history.csv b/previous_single_task/GraspCylinder-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..2cdbbfa8c9ad30bc11160174e2e51129ccfbdbfa --- /dev/null +++ b/previous_single_task/GraspCylinder-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,45,0.9,24.33076073028518,True +10000,50,45,0.9,24.851561762759992,True +15000,50,47,0.94,25.653659981940546,True +20000,50,45,0.9,24.57017223808127,True +25000,50,46,0.92,25.68929285959772,True +30000,50,47,0.94,26.218032003741765,True +35000,50,47,0.94,25.841893786347146,True +40000,50,45,0.9,25.292166099923634,True diff --git a/previous_single_task/GraspCylinder-v1/lora_step_0015000.npz b/previous_single_task/GraspCylinder-v1/lora_step_0015000.npz new file mode 100644 index 0000000000000000000000000000000000000000..d2516164580c98e54ee6efefa6dbf6fe001107b5 --- /dev/null +++ b/previous_single_task/GraspCylinder-v1/lora_step_0015000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8df817057bb6ac221d574592a7437a1607aae1853e0350e3426507215748d225 +size 1867844866 diff --git a/previous_single_task/GraspCylinder-v1/metadata.json b/previous_single_task/GraspCylinder-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..abb6f3a63fec5dc268ce65a292810337b625d4e4 --- /dev/null +++ b/previous_single_task/GraspCylinder-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "GraspCylinder-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0015000.npz", + "checkpoint_sha256": "8df817057bb6ac221d574592a7437a1607aae1853e0350e3426507215748d225", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 15000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "f2079b048188d674e79b6f787d5ebe3a6e9de30fdfe694c7d013ee339b9cd8e8", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_GraspCylinder-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_1/grasp_cylinder.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/GraspCylinder-v1/norm_stats.json b/previous_single_task/GraspCylinder-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..9eeb22d9115ebcfbeb4ba2ef9ba804de7dcb0012 --- /dev/null +++ b/previous_single_task/GraspCylinder-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.01269422098994255, + -0.0066573903895914555, + 0.021332062780857086, + 0.0, + 0.0, + 0.0, + -0.009019139222800732, + 0.10273551940917969, + 0.08068190515041351, + 0.06767021864652634, + -0.002929937094449997, + 0.0757085308432579, + 0.0777587890625, + 0.06853114813566208, + -0.00005274502473184839, + 0.06694968789815903, + 0.0646234080195427, + 0.05693183094263077, + -0.0032970686443150043, + -0.0007048468687571585, + 0.15879392623901367, + 0.09889224171638489, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.05508832260966301, + 0.03451097384095192, + 0.061533283442258835, + 0.0, + 0.0, + 0.0, + 0.01166828628629446, + 0.09944471716880798, + 0.08823512494564056, + 0.08532287925481796, + 0.006191034335643053, + 0.08737068623304367, + 0.08857345581054688, + 0.08602859824895859, + 0.0028655026108026505, + 0.08540559560060501, + 0.08501263707876205, + 0.08629713207483292, + 0.018372271209955215, + 0.008411682210862637, + 0.13826939463615417, + 0.09687664359807968, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -0.03357360384333879, + 0.0, + 0.0, + 0.0, + -0.020345867547206582, + 0.0, + 0.0, + 0.0, + -0.006592770796269179, + 0.0, + 0.0, + 0.0, + -0.03342294429540634, + -0.018650883591175078, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + 0.0005799511522054657, + 0.3666337996721268, + 0.3666053652763367, + 0.3667560145735741, + 0.0039945648731663855, + 0.3666337156295776, + 0.36668333839774137, + 0.3667560145735741, + 0.008343444367870689, + 0.3666336315870285, + 0.36668333839774137, + 0.3667560145735741, + 0.05766023803949355, + 0.022922335872054105, + 0.3667255087137222, + 0.36675609865188596, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/GraspCylinder-v1/source_audit.json b/previous_single_task/GraspCylinder-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..51baa24ccaa9da8eb96dcce3b3a8b2fec1c23c49 --- /dev/null +++ b/previous_single_task/GraspCylinder-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "GraspCylinder-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.9", + "peak_step": "10000", + "final": "0.85", + "start": "2026-08-21 13:36:54", + "eval_mtime": "2026-08-24 20:42:00.132909", + "source_mtime": "2026-05-19 02:30:46.162825", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_1/grasp_cylinder.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_GraspCylinder-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_GraspCylinder-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_GraspCylinder-v1/lora_params/lora_step_0010000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/GraspTwoBlocks-v1/eval_result.json b/previous_single_task/GraspTwoBlocks-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..41fc3b4d43530519f1732a18163cc810b6a3c1ef --- /dev/null +++ b/previous_single_task/GraspTwoBlocks-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "GraspTwoBlocks-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_dex", + "step": 5000, + "episodes": 50, + "successes": 36, + "success_rate": 0.72, + "return_mean": 199.64137156695128, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/GraspTwoBlocks-v1/evaluation_history.csv b/previous_single_task/GraspTwoBlocks-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..545b6e2d5b395d1a3129b9fc03c04b03ed425bbb --- /dev/null +++ b/previous_single_task/GraspTwoBlocks-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,36,0.72,199.64137156695128,True +10000,50,29,0.58,215.0543767988682,True +15000,50,27,0.54,222.43316952705382,True +20000,50,28,0.56,216.99497130930425,True +25000,50,26,0.52,224.06823341548443,True +30000,50,28,0.56,224.0523162674904,True +35000,50,24,0.48,226.63711440443993,True +40000,50,22,0.44,236.3215113723278,True diff --git a/previous_single_task/GraspTwoBlocks-v1/lora_step_0005000.npz b/previous_single_task/GraspTwoBlocks-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..eac50a90ee744cd2b42d8963661644d43632e523 --- /dev/null +++ b/previous_single_task/GraspTwoBlocks-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:84de21676c4ded27f090d19e152584578a07ce0de6816d732e84f967a9378831 +size 1867844866 diff --git a/previous_single_task/GraspTwoBlocks-v1/metadata.json b/previous_single_task/GraspTwoBlocks-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..b59dedc01441626414e5958629cb1f5b121c1ef9 --- /dev/null +++ b/previous_single_task/GraspTwoBlocks-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "GraspTwoBlocks-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "84de21676c4ded27f090d19e152584578a07ce0de6816d732e84f967a9378831", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "ad947570ec8077e915eb526a287d8b9682b933e0db5b2d746b8f8edcd067e082", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_GraspTwoBlocks-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_2/grasp_blocks.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/GraspTwoBlocks-v1/norm_stats.json b/previous_single_task/GraspTwoBlocks-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..2074e3ad032c8ccfcf1e56a510f7142acc90f263 --- /dev/null +++ b/previous_single_task/GraspTwoBlocks-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.00238017737865448, + 0.015725575387477875, + -0.013917149975895882, + 0.0, + 0.0, + 0.0, + -0.001966198207810521, + 0.08892150968313217, + 0.07670991122722626, + 0.0736687108874321, + -0.00001394227183482144, + 0.07287771999835968, + 0.07275298982858658, + 0.07266338169574738, + 2.5463150450377725e-6, + 0.0725989118218422, + 0.07261361181735992, + 0.07261724770069122, + 0.04667350649833679, + -0.0021363324485719204, + 0.09634091705083847, + 0.08039785176515579, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.02009527198970318, + 0.035784896463155746, + 0.04929351434111595, + 0.0, + 0.0, + 0.0, + 0.0031324259471148252, + 0.10058821737766266, + 0.09844592213630676, + 0.09855419397354126, + 0.00003713518526637927, + 0.09869764000177383, + 0.09872475266456604, + 0.0987408310174942, + 4.383491159387631e-6, + 0.09873480349779129, + 0.09874729067087173, + 0.098750039935112, + 0.20155754685401917, + 0.0036829456221312284, + 0.10357669740915298, + 0.0990101620554924, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.0512400007635355, + -0.0011200000166892987, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -0.008519075423780305, + 0.0, + 0.0, + 0.0, + -0.00014517796975851524, + 0.0, + 0.0, + 0.0, + -4.6837058289384e-6, + 0.0, + 0.0, + 0.0, + -0.08642813715934754, + -0.011009610752016306, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + 0.000017206310843903214, + 0.3780587346494198, + 0.37822039650082584, + 0.37837801503539087, + 0.0000409903910011053, + 0.37797819432020185, + 0.37822004967331885, + 0.37837801503539087, + 0.00001936637419930776, + 0.37805362105369567, + 0.37821967394351963, + 0.37837801503539087, + 0.7898363893032073, + 0.0008772966414224356, + 0.37835580796003343, + 0.3783000439405441, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/GraspTwoBlocks-v1/source_audit.json b/previous_single_task/GraspTwoBlocks-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..25d96151638b5623d19a3c745e9d8bf00940eb1e --- /dev/null +++ b/previous_single_task/GraspTwoBlocks-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "GraspTwoBlocks-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.6", + "peak_step": "10000", + "final": "0.4", + "start": "2026-08-21 13:36:57", + "eval_mtime": "2026-08-24 19:56:13.810243", + "source_mtime": "2026-04-03 14:28:20.432960", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_2/grasp_blocks.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_GraspTwoBlocks-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_GraspTwoBlocks-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_GraspTwoBlocks-v1/lora_params/lora_step_0010000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/HammerNail-v1/eval_result.json b/previous_single_task/HammerNail-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..58e1b395bd96a2ce4ab984d2e9ffa99a8379e1f6 --- /dev/null +++ b/previous_single_task/HammerNail-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "HammerNail-v1", + "level": "level_3", + "suite": "rigid_body/single_arm_dex", + "step": 35000, + "episodes": 50, + "successes": 6, + "success_rate": 0.12, + "return_mean": null, + "return_mean_finite": false, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/HammerNail-v1/evaluation_history.csv b/previous_single_task/HammerNail-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..9f2f96c11f50858f0a068a6ad6de583ea6f85c8d --- /dev/null +++ b/previous_single_task/HammerNail-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,2,0.04,-58.59067942216993,True +10000,50,3,0.06,-58.978516302481296,True +15000,50,5,0.1,,False +20000,50,3,0.06,-58.93580446228385,True +25000,50,5,0.1,-53.304630602933464,True +30000,50,4,0.08,-57.3198606460169,True +35000,50,6,0.12,,False +40000,50,5,0.1,-59.44200713545084,True diff --git a/previous_single_task/HammerNail-v1/lora_step_0035000.npz b/previous_single_task/HammerNail-v1/lora_step_0035000.npz new file mode 100644 index 0000000000000000000000000000000000000000..ac1dd88fafd0ae84caa76fd9058d548acf25c5d9 --- /dev/null +++ b/previous_single_task/HammerNail-v1/lora_step_0035000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a26699d0543203305445941f7008a0c0ed01afa2b451ffa3850d54fbc45f4dc9 +size 1867844866 diff --git a/previous_single_task/HammerNail-v1/metadata.json b/previous_single_task/HammerNail-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..0a79de81c6976476dbeda3ca572055f284c14fc2 --- /dev/null +++ b/previous_single_task/HammerNail-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "HammerNail-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0035000.npz", + "checkpoint_sha256": "a26699d0543203305445941f7008a0c0ed01afa2b451ffa3850d54fbc45f4dc9", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 35000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "34e43d9863b241d7be3eff75ee0afa327f9a1cb780908ce4216b7fd67b8a3d41", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_HammerNail-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_3/hammer_nail.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/HammerNail-v1/norm_stats.json b/previous_single_task/HammerNail-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..c0f377908550f14de4f796aee93cf29ff1258c69 --- /dev/null +++ b/previous_single_task/HammerNail-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.006555285770446062, + -0.0037326645106077194, + -0.010760940611362457, + 0.0, + 0.0, + 0.0, + -0.001675690058618784, + 0.06119761988520622, + 0.04357372224330902, + 0.03664681687951088, + -0.0023865150287747383, + 0.05010617896914482, + 0.04440344497561455, + 0.038026418536901474, + -1.5338751779836457e-7, + 0.029759736731648445, + 0.017861386761069298, + 0.022562382742762566, + -0.0037031767424196005, + -0.0009390070335939527, + 0.05004677549004555, + 0.05128944292664528, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.025101438164711, + 0.06647323817014694, + 0.052664127200841904, + 0.0, + 0.0, + 0.0, + 0.006205757148563862, + 0.07040344178676605, + 0.06406784057617188, + 0.06235116720199585, + 0.0032825409434735775, + 0.06308598071336746, + 0.06274909526109695, + 0.062231313437223434, + 3.3237874959013425e-6, + 0.06577307730913162, + 0.06898628920316696, + 0.06580361723899841, + 0.13900689780712128, + 0.007676869165152311, + 0.09690173715353012, + 0.07151760160923004, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -0.022313025482557713, + -0.00017257988452911398, + -0.030626978927850722, + 0.0, + -0.011763168750703334, + -0.0001377306222915646, + -0.018026646995544432, + 0.0, + -7.990696003253107e-6, + -0.0067992453932762195, + -0.07525836129784584, + -0.010886233055591584, + -0.10012795994281769, + -0.008215600018575787, + -0.12151465475559234, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0025600000381469784, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + 0.0029349769433960303, + 0.33406822216510773, + 0.33409453688263896, + 0.33407403880357744, + 0.002489850179851056, + 0.33404444481134415, + 0.3340530557870865, + 0.33407403880357744, + 0.000014150793170847461, + 0.3340574086785316, + 0.33402507735490794, + 0.33404182809591293, + 0.7708965041041373, + 0.021673973045125604, + 0.3340601997375488, + 0.3340740643322468, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/HammerNail-v1/source_audit.json b/previous_single_task/HammerNail-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..cfca331d48444db7a0634547c0c52eb797d2433b --- /dev/null +++ b/previous_single_task/HammerNail-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "HammerNail-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.15", + "peak_step": "10000", + "final": "0.15", + "start": "2026-08-22 06:41:04", + "eval_mtime": "2026-08-25 05:38:05.920484", + "source_mtime": "2026-04-03 14:28:20.436960", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_3/hammer_nail.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_HammerNail-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_HammerNail-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_HammerNail-v1/lora_params/lora_step_0010000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/HammerNailOut-v1/eval_result.json b/previous_single_task/HammerNailOut-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..550856ab574326e29e3011394e09fdd9a3990828 --- /dev/null +++ b/previous_single_task/HammerNailOut-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "HammerNailOut-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_dex", + "step": 40000, + "episodes": 50, + "successes": 22, + "success_rate": 0.44, + "return_mean": -25.41208131439984, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/HammerNailOut-v1/evaluation_history.csv b/previous_single_task/HammerNailOut-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..e6a883efdacb2efdcf7da3fcd2fc070601c3087f --- /dev/null +++ b/previous_single_task/HammerNailOut-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,12,0.24,-27.080348949953915,True +10000,50,15,0.3,-27.273450547121467,True +15000,50,12,0.24,-27.64660687096417,True +20000,50,13,0.26,-22.220520413033665,True +25000,50,19,0.38,-26.580098412930965,True +30000,50,14,0.28,6.585363882668316,True +35000,50,19,0.38,-26.564064246863126,True +40000,50,22,0.44,-25.41208131439984,True diff --git a/previous_single_task/HammerNailOut-v1/lora_step_0040000.npz b/previous_single_task/HammerNailOut-v1/lora_step_0040000.npz new file mode 100644 index 0000000000000000000000000000000000000000..9e861d40c7dd12636c3cd599d39ceb8ef42f8180 --- /dev/null +++ b/previous_single_task/HammerNailOut-v1/lora_step_0040000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8dc1babeb6ad28356e7a4fedc58122db41c17d4564e0ac1d95cd7c8fbe5b228a +size 1867844866 diff --git a/previous_single_task/HammerNailOut-v1/metadata.json b/previous_single_task/HammerNailOut-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..13b3a2fcab85a58d7973f723dca20947de958a72 --- /dev/null +++ b/previous_single_task/HammerNailOut-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "HammerNailOut-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0040000.npz", + "checkpoint_sha256": "8dc1babeb6ad28356e7a4fedc58122db41c17d4564e0ac1d95cd7c8fbe5b228a", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 40000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "75abd49268917b7b79fe2db3706d6ec9fe96e59da2a9629df4cb34d7278badf8", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_HammerNailOut-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_2/hammer_nail_out.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/HammerNailOut-v1/norm_stats.json b/previous_single_task/HammerNailOut-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..688cec7b0558eec848cf451ecb3800df3f0aa473 --- /dev/null +++ b/previous_single_task/HammerNailOut-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.024288106709718704, + 0.008499762043356895, + 0.0038325460627675056, + 0.0, + -0.0200535599142313, + 0.0, + 0.0012773207854479551, + 0.06056177243590355, + 0.04161810874938965, + 0.032697539776563644, + 0.0014790277928113937, + 0.054655466228723526, + 0.04019984230399132, + 0.03279896825551987, + 0.002965725027024746, + 0.05886402353644371, + 0.04675621911883354, + 0.03702074661850929, + 0.029440047219395638, + -0.0001464181550545618, + 0.004372219555079937, + 0.0478844977915287, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.07655350118875504, + 0.07898620516061783, + 0.10481207817792892, + 0.0, + 0.074925996363163, + 0.0, + 0.00194578489754349, + 0.11833357810974121, + 0.07609831541776657, + 0.0830855518579483, + 0.0019021761836484075, + 0.12931303679943085, + 0.07448917627334595, + 0.10299348086118698, + 0.0038166309241205454, + 0.12822718918323517, + 0.07330361753702164, + 0.10148274153470993, + 0.06841784715652466, + 0.001167972106486559, + 0.0038105978164821863, + 0.11251702904701233, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.15000000596046448, + -0.15000000596046448, + -0.15000000596046448, + 0.0, + -0.30000001192092896, + 0.0, + -0.003085346394404769, + -0.05145334019660949, + 0.0, + -0.12995328627228736, + -0.000010411956027382985, + -0.04421884317398071, + -0.000014762761566089466, + -0.11800448690652847, + -9.626720821324852e-6, + -0.041226530051231385, + -0.00001481020990468096, + -0.1105788323879242, + -0.0072514022767543795, + -0.0034454012257046996, + -0.000026685204727982637, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.14994000595808032, + 0.14994000595808032, + 0.14994000595808032, + 0.0, + -0.00006000000238415293, + 0.0, + 0.007373464152589439, + 0.601715640115738, + 0.4083397536277771, + 0.36534411522150034, + 0.006399823136162012, + 0.6551538647651672, + 0.3891952285795007, + 0.5158953080415726, + 0.01409780441028779, + 0.6552421250343323, + 0.38919508866795804, + 0.515770644390583, + 0.3651093305528164, + 0.0020469528860878198, + 0.013497403235844103, + 0.6017761093139649, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/HammerNailOut-v1/source_audit.json b/previous_single_task/HammerNailOut-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..19c4a44b632241904c017026c14c57670ae89a7a --- /dev/null +++ b/previous_single_task/HammerNailOut-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "HammerNailOut-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.55", + "peak_step": "35000", + "final": "0.35", + "start": "2026-08-21 14:50:34", + "eval_mtime": "2026-08-24 14:42:04.514508", + "source_mtime": "2026-06-23 14:55:40.742644", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_2/hammer_nail_out.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_HammerNailOut-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_HammerNailOut-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_HammerNailOut-v1/lora_params/lora_step_0035000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/HangMug-v1/eval_result.json b/previous_single_task/HangMug-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..59c911fb110de77d386e7053f85bc9c10bf91ab5 --- /dev/null +++ b/previous_single_task/HangMug-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "HangMug-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_gripper", + "step": 10000, + "episodes": 50, + "successes": 35, + "success_rate": 0.7, + "return_mean": 31.07875474691391, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/HangMug-v1/evaluation_history.csv b/previous_single_task/HangMug-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..33f1360a5b7f8affa3763a426dd67ac153d08ac4 --- /dev/null +++ b/previous_single_task/HangMug-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,34,0.68,30.998696595430374,True +10000,50,35,0.7,31.07875474691391,True +15000,50,30,0.6,32.206367956399916,True +20000,50,28,0.56,31.178596129417418,True +25000,50,29,0.58,34.236433204412464,True +30000,50,31,0.62,28.8294646692276,True +35000,50,31,0.62,33.019806396961215,True +40000,50,29,0.58,31.526209642887114,True diff --git a/previous_single_task/HangMug-v1/lora_step_0010000.npz b/previous_single_task/HangMug-v1/lora_step_0010000.npz new file mode 100644 index 0000000000000000000000000000000000000000..3054ee09062bc6452aa72a753a3f794993b5eaba --- /dev/null +++ b/previous_single_task/HangMug-v1/lora_step_0010000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:7ad647a1261390f68a70fa7320c3238a05aec39db9d37d8a34be76c63ee1c8ed +size 1867844866 diff --git a/previous_single_task/HangMug-v1/metadata.json b/previous_single_task/HangMug-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..d6e54c4c699606e919f64f2f68d0675c63dc8808 --- /dev/null +++ b/previous_single_task/HangMug-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "HangMug-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0010000.npz", + "checkpoint_sha256": "7ad647a1261390f68a70fa7320c3238a05aec39db9d37d8a34be76c63ee1c8ed", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 10000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "8e89b4c6d2a30eb88abce3af55cef3fcafa8592f738e64eb0903b9e43838770d", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_HangMug-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_2/hang_mug.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/HangMug-v1/norm_stats.json b/previous_single_task/HangMug-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..f74980d6243adcc8dd6650eddb484df8518154b5 --- /dev/null +++ b/previous_single_task/HangMug-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0030757796484977007, + 0.003326000180095434, + -0.011506158858537674, + 0.0, + 0.0, + 0.0, + 0.06057959794998169, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.03953126445412636, + 0.061851851642131805, + 0.06229623034596443, + 0.0, + 0.0, + 0.0, + 0.9981634020805359, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/HangMug-v1/source_audit.json b/previous_single_task/HangMug-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..3b945ce94f7af37d5fe846893207a2f7165853f8 --- /dev/null +++ b/previous_single_task/HangMug-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "HangMug-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.8", + "peak_step": "5000", + "final": "0.7", + "start": "2026-08-22 05:49:04", + "eval_mtime": "2026-08-25 04:45:28.143601", + "source_mtime": "2026-06-05 12:33:58.596056", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_2/hang_mug.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_HangMug-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_HangMug-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_HangMug-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/HitBilliard-v1/eval_result.json b/previous_single_task/HitBilliard-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..f805d5b652705837a51b5366b28748847b26f5f6 --- /dev/null +++ b/previous_single_task/HitBilliard-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "HitBilliard-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": 852.1505782365799, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/HitBilliard-v1/evaluation_history.csv b/previous_single_task/HitBilliard-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..d1e880e2f9b630772e7a8ae8db52746a08bb3869 --- /dev/null +++ b/previous_single_task/HitBilliard-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,852.1505782365799,True +10000,50,0,0.0,838.0057423710823,True +15000,50,0,0.0,817.1832637572288,True +20000,50,0,0.0,819.156718518734,True +25000,50,0,0.0,763.028509516716,True +30000,50,0,0.0,806.5827033662796,True +35000,50,0,0.0,795.0292140698433,True +40000,50,0,0.0,794.3330316150189,True diff --git a/previous_single_task/HitBilliard-v1/lora_step_0005000.npz b/previous_single_task/HitBilliard-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..d9a242d27895d07781f2eba98758a4fa9a8a9d6f --- /dev/null +++ b/previous_single_task/HitBilliard-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:f047f94f21e69846042a12ec38e75c3bb6e4467f9bfb9eeba39d43db46800748 +size 1867844866 diff --git a/previous_single_task/HitBilliard-v1/metadata.json b/previous_single_task/HitBilliard-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..601a3bdb7a565ff4c4f826e6d80792127aa6dc93 --- /dev/null +++ b/previous_single_task/HitBilliard-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "HitBilliard-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "f047f94f21e69846042a12ec38e75c3bb6e4467f9bfb9eeba39d43db46800748", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "4ac9cff682a3153e28bd4efa489278bd12185190e381067767de11f2d2b2fa66", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_HitBilliard-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_4/hit_billiard.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/HitBilliard-v1/norm_stats.json b/previous_single_task/HitBilliard-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..ec80b8c3c6a370057950b1e3f5535b4db88a5c4b --- /dev/null +++ b/previous_single_task/HitBilliard-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.8400408625602722, + -1.0517613887786865, + 0.881274402141571, + 0.005229406990110874, + 0.008887900039553642, + 0.004773963242769241, + 0.000148236911627464, + 0.00020079160458408296, + 0.006762456148862839, + 0.15849162638187408, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.420288622379303, + 0.43009844422340393, + 0.7648270726203918, + 0.02650088258087635, + 0.05310366302728653, + 0.03890008106827736, + 0.0016214856877923012, + 0.002320434432476759, + 0.05172143131494522, + 0.9873602986335754, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.24982968577742576, + -1.6061335802078247, + -0.0006258279900066555, + -0.06375249525010586, + -0.004165830731391907, + -0.11999999731779099, + -0.0018714655622839964, + -0.0005605563026852911, + -0.019713810998201353, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 1.4843828691244125, + -0.44199888386130337, + 1.5801709339573164, + 0.11995646438896657, + 0.3896701721549034, + 0.11995199731886386, + 0.010024029962718485, + 0.009949721314106135, + 0.3498693069398403, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/HitBilliard-v1/source_audit.json b/previous_single_task/HitBilliard-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..b3caaeb20a375f0be1b2a3039fb992320012991c --- /dev/null +++ b/previous_single_task/HitBilliard-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "HitBilliard-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-21 13:36:59", + "eval_mtime": "2026-08-25 11:38:20.728025", + "source_mtime": "2026-05-14 15:55:51.776870", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_4/hit_billiard.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_HitBilliard-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_HitBilliard-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_HitBilliard-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/HitBox-v1/eval_result.json b/previous_single_task/HitBox-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..a8b6d1b8b556b966f82a4688bd11f62cfc58c8bd --- /dev/null +++ b/previous_single_task/HitBox-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "HitBox-v1", + "level": "level_3", + "suite": "deformable/single_arm_dex", + "step": 25000, + "episodes": 50, + "successes": 3, + "success_rate": 0.06, + "return_mean": 351.1648123627901, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/HitBox-v1/evaluation_history.csv b/previous_single_task/HitBox-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..20b386009c4eb4ec9a19791377eb7518bc777b15 --- /dev/null +++ b/previous_single_task/HitBox-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,341.86592402637007,True +10000,50,0,0.0,302.5687988591194,True +15000,50,2,0.04,291.51658733904367,True +20000,50,0,0.0,320.2418014472723,True +25000,50,3,0.06,351.1648123627901,True +30000,50,0,0.0,316.49283266067505,True +35000,50,2,0.04,337.8158902138472,True +40000,50,0,0.0,330.6795945161581,True diff --git a/previous_single_task/HitBox-v1/lora_step_0025000.npz b/previous_single_task/HitBox-v1/lora_step_0025000.npz new file mode 100644 index 0000000000000000000000000000000000000000..daeadb69aa8670364e3d24758700e431ac534b2f --- /dev/null +++ b/previous_single_task/HitBox-v1/lora_step_0025000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:fc4b6cdf6398e6b7222b8a9ba80630fbd262fb2de923829c55aa3aa1c2df27c0 +size 1867844866 diff --git a/previous_single_task/HitBox-v1/metadata.json b/previous_single_task/HitBox-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..54035231b77e36358a16b5b1c2d0cb5935636c9e --- /dev/null +++ b/previous_single_task/HitBox-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "HitBox-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0025000.npz", + "checkpoint_sha256": "fc4b6cdf6398e6b7222b8a9ba80630fbd262fb2de923829c55aa3aa1c2df27c0", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 25000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "0052614e3f4704f916c6330e579d0bd644b2357d43675c112d5e292358eb058e", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_HitBox-v1", + "source_path": "maniladder/envs/deformable/single_arm_dex/level_3/hit_box.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/HitBox-v1/norm_stats.json b/previous_single_task/HitBox-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..cc158ea380b81e500f5dc50be171571c435ac66e --- /dev/null +++ b/previous_single_task/HitBox-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.0012287833960726857, + -0.029503270983695984, + 0.008164510130882263, + -0.00585744995623827, + 0.14018790423870087, + 0.021037831902503967, + -0.013520436361432076, + 0.0366671085357666, + 0.03456810489296913, + 0.015430735424160957, + -0.0015914862742647529, + 0.05095171555876732, + 0.0386224128305912, + 0.014523351565003395, + -0.0006491797976195812, + 0.003343023592606187, + 0.0034664920531213284, + 0.0014302466297522187, + -0.051232948899269104, + -0.0008012124453671277, + 0.14422443509101868, + 0.06877566128969193, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.06433120369911194, + 0.10278023779392242, + 0.06845595687627792, + 0.027390440925955772, + 0.1997244507074356, + 0.061545442789793015, + 0.011029168963432312, + 0.14645588397979736, + 0.14073088765144348, + 0.14835646748542786, + 0.0017755161970853806, + 0.1395437866449356, + 0.1411241888999939, + 0.15049180388450623, + 0.0012282211100682616, + 0.1604209989309311, + 0.15993395447731018, + 0.15915273129940033, + 0.14358006417751312, + 0.017797240987420082, + 0.1387522667646408, + 0.13814061880111694, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.30000001192092896, + -0.30000001192092896, + -0.26278911110162734, + -0.14659348520934581, + -0.15425190902352334, + -0.09372117279171943, + -0.04452939372985163, + -0.76554475158453, + -0.7446260102391242, + -0.8218840342342855, + -0.005579686666000634, + -0.7002384806871415, + -0.7371376861333847, + -0.8338289870679378, + -0.005536186293276841, + -0.9098625450134277, + -0.9098790689826012, + -0.9098942260146141, + -0.7705135982036591, + -0.083118549644202, + -0.3757261652290821, + -0.6408816885530948, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.12527999441623688, + 0.3499199977636338, + 0.12956629656553276, + 0.034993394476175305, + 0.9685752451837062, + 0.37022258293628696, + -0.000013018892127306114, + 0.35664716824889187, + 0.3563662371635439, + 0.35652255555987344, + 0.001226253993436694, + 0.3563675467967986, + 0.3565967492580415, + 0.3564969314575195, + 8.66433713235873e-6, + 0.35614902973175044, + 0.3565329371511936, + 0.3565393992960453, + -0.00020612990856161417, + 0.012537214917689568, + 0.3567023051381112, + 0.35656309912204764, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/HitBox-v1/source_audit.json b/previous_single_task/HitBox-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..594351a057960b699f169991fc0da09e41c4077f --- /dev/null +++ b/previous_single_task/HitBox-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "HitBox-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.05", + "peak_step": "20000", + "final": "0.05", + "start": "2026-09-04 06:04:25", + "eval_mtime": "2026-09-07 08:10:44.286883", + "source_mtime": "2026-08-29 13:22:56.217699", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/single_arm_dex/level_3/hit_box.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_HitBox-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_HitBox-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_HitBox-v1/lora_params/lora_step_0020000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/KitAssembly-v1/eval_result.json b/previous_single_task/KitAssembly-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..27fdf973568f80e7f9ef378ff011f7edcf9839fb --- /dev/null +++ b/previous_single_task/KitAssembly-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "KitAssembly-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_gripper", + "step": 15000, + "episodes": 50, + "successes": 3, + "success_rate": 0.06, + "return_mean": 423.16, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/KitAssembly-v1/evaluation_history.csv b/previous_single_task/KitAssembly-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..58b3a11ccafd7e0be4db63ed90d87bda846cd30b --- /dev/null +++ b/previous_single_task/KitAssembly-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,417.955,True +10000,50,1,0.02,322.37,True +15000,50,3,0.06,423.16,True +20000,50,3,0.06,312.665,True +25000,50,0,0.0,134.07,True +30000,50,0,0.0,232.83,True +35000,50,0,0.0,324.45,True +40000,50,0,0.0,188.325,True diff --git a/previous_single_task/KitAssembly-v1/lora_step_0015000.npz b/previous_single_task/KitAssembly-v1/lora_step_0015000.npz new file mode 100644 index 0000000000000000000000000000000000000000..2296a51c35c18f9a3f7eb695a35b6d3d6ab69de7 --- /dev/null +++ b/previous_single_task/KitAssembly-v1/lora_step_0015000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2e78e125fcb2aacf7a1c59973868de4839dc8bf7c6fdfa649b5ebde9b3841189 +size 1867844866 diff --git a/previous_single_task/KitAssembly-v1/metadata.json b/previous_single_task/KitAssembly-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..23abae61aca43d014bc3c9afafc95408227c0e50 --- /dev/null +++ b/previous_single_task/KitAssembly-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "KitAssembly-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0015000.npz", + "checkpoint_sha256": "2e78e125fcb2aacf7a1c59973868de4839dc8bf7c6fdfa649b5ebde9b3841189", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 15000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "3f256a8417057bab0e12ca605b786a4d107940b929fc7bf77217a77fc8af4fb2", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_KitAssembly-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_2/kit_assembly.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/KitAssembly-v1/norm_stats.json b/previous_single_task/KitAssembly-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..e80febc632a764fb1ba8bc434740419159478035 --- /dev/null +++ b/previous_single_task/KitAssembly-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.002305446658283472, + -0.0008195766713470221, + -0.0019183352123945951, + 0.0, + 0.0, + 0.0, + -0.022135412320494652, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.026095613837242126, + 0.04844064638018608, + 0.07444571703672409, + 0.0, + 0.0, + 0.0, + 0.9997549653053284, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/KitAssembly-v1/source_audit.json b/previous_single_task/KitAssembly-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..1d67c2e44df40802191c822b204c9e3b23426621 --- /dev/null +++ b/previous_single_task/KitAssembly-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "KitAssembly-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.1", + "peak_step": "15000", + "final": "0.0", + "start": "2026-08-22 06:21:10", + "eval_mtime": "2026-08-25 15:39:15.193506", + "source_mtime": "2026-05-15 05:51:37.924556", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_2/kit_assembly.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_KitAssembly-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_KitAssembly-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_KitAssembly-v1/lora_params/lora_step_0015000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/LiftBucket-v1/eval_result.json b/previous_single_task/LiftBucket-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..2d5eb8403f4d1da1b4fbf518164bd833fe636aaf --- /dev/null +++ b/previous_single_task/LiftBucket-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "LiftBucket-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_dex", + "step": 40000, + "episodes": 50, + "successes": 48, + "success_rate": 0.96, + "return_mean": 96.80372586607933, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/LiftBucket-v1/evaluation_history.csv b/previous_single_task/LiftBucket-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..1cce4943ca45691245b8c578e40b8d0d011b0470 --- /dev/null +++ b/previous_single_task/LiftBucket-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,10,0.2,163.83099677085877,True +10000,50,47,0.94,98.44282671809196,True +15000,50,42,0.84,106.98082929849625,True +20000,50,43,0.86,106.26126730442047,True +25000,50,46,0.92,103.61563530683517,True +30000,50,42,0.84,108.06760128974915,True +35000,50,46,0.92,100.6508888375759,True +40000,50,48,0.96,96.80372586607933,True diff --git a/previous_single_task/LiftBucket-v1/lora_step_0040000.npz b/previous_single_task/LiftBucket-v1/lora_step_0040000.npz new file mode 100644 index 0000000000000000000000000000000000000000..e11f46f767c1e1a7c58b69b36ed2ebd31122bc88 --- /dev/null +++ b/previous_single_task/LiftBucket-v1/lora_step_0040000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5040809ab7b421527b71dce655b290016467761ca5460a43a32fa2121d2519ee +size 1867844866 diff --git a/previous_single_task/LiftBucket-v1/metadata.json b/previous_single_task/LiftBucket-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..28f3c46412a10c7ebe68e1621920955cc78c8632 --- /dev/null +++ b/previous_single_task/LiftBucket-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "LiftBucket-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0040000.npz", + "checkpoint_sha256": "5040809ab7b421527b71dce655b290016467761ca5460a43a32fa2121d2519ee", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 40000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "12733894b07e6079c08d75c7c6adb9c91aa5f776a0a1b2b5aea35f80f76003a4", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_LiftBucket-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_1/lift_bucket.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/LiftBucket-v1/norm_stats.json b/previous_single_task/LiftBucket-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..862a2cea495e76299143b8363bb88a6000b34afa --- /dev/null +++ b/previous_single_task/LiftBucket-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0012979306047782302, + 0.021129116415977478, + 0.04566526040434837, + 0.0, + 0.0, + 0.0, + 0.0016858505550771952, + 0.08572909235954285, + 0.07648036628961563, + 0.07442978024482727, + 0.0021127350628376007, + 0.0794641301035881, + 0.0841398686170578, + 0.07957855612039566, + -0.00039698564796708524, + 0.07739431411027908, + 0.07881546765565872, + 0.0762811154127121, + 0.05753793939948082, + 0.009896278381347656, + 0.12255153805017471, + 0.10187563300132751, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.012354454956948757, + 0.09748285263776779, + 0.04781784117221832, + 0.0, + 0.0, + 0.0, + 0.0024150866083800793, + 0.09675626456737518, + 0.09618455916643143, + 0.09649045020341873, + 0.0033751942683011293, + 0.09546402841806412, + 0.09507164359092712, + 0.09548141062259674, + 0.0008508278988301754, + 0.09618814289569855, + 0.09555505216121674, + 0.09600301831960678, + 0.0749504491686821, + 0.012045721523463726, + 0.1069006472826004, + 0.09864558279514313, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.005520000082254409, + -0.10000000149011612, + -0.023520000350475315, + 0.0, + 0.0, + 0.0, + -0.001989211613498628, + 0.0, + 0.0, + 0.0, + -0.00003869185408257181, + 0.0, + 0.0, + 0.0, + -0.0032679265784681774, + 0.0, + 0.0, + 0.0, + 0.0, + -0.000043735237792134223, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + 0.0057658415069803595, + 0.3783075311779976, + 0.37830180734395985, + 0.37837801503539087, + 0.010274560034905152, + 0.37830796480178835, + 0.37830186516046527, + 0.37837801503539087, + 0.00015807713564718138, + 0.3783083695173264, + 0.37830189406871795, + 0.37837801503539087, + 0.2910259133577347, + 0.032049501176178455, + 0.3782650071382523, + 0.378299754858017, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/LiftBucket-v1/source_audit.json b/previous_single_task/LiftBucket-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..b210e774d3da3bd2dff5eedefcde7c690b944e15 --- /dev/null +++ b/previous_single_task/LiftBucket-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "LiftBucket-v1", + "category": "reuse", + "level": "level_1", + "peak": "1.0", + "peak_step": "10000", + "final": "0.85", + "start": "2026-08-21 13:37:01", + "eval_mtime": "2026-08-25 00:37:32.259519", + "source_mtime": "2026-02-18 06:30:34.264665", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_1/lift_bucket.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_LiftBucket-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_LiftBucket-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_LiftBucket-v1/lora_params/lora_step_0010000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/LiquidMeasuring-v1/eval_result.json b/previous_single_task/LiquidMeasuring-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..b8a7f9151ddbce2f4edb4279cb21f59a533b31a9 --- /dev/null +++ b/previous_single_task/LiquidMeasuring-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "LiquidMeasuring-v1", + "level": "level_3", + "suite": "deformable/single_arm_gripper", + "step": 35000, + "episodes": 50, + "successes": 16, + "success_rate": 0.32, + "return_mean": 95.25492954222486, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/LiquidMeasuring-v1/evaluation_history.csv b/previous_single_task/LiquidMeasuring-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..c442ce7a37f2cc4807b5d3f866f373e4c3ccbae4 --- /dev/null +++ b/previous_single_task/LiquidMeasuring-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,9,0.18,124.16856057690921,True +10000,50,8,0.16,,False +15000,50,14,0.28,97.86243267653975,True +20000,50,9,0.18,111.7851439455105,True +25000,50,9,0.18,118.19692628441379,True +30000,50,10,0.2,117.9736181830382,True +35000,50,16,0.32,95.25492954222486,True +40000,50,13,0.26,118.64093734898138,True diff --git a/previous_single_task/LiquidMeasuring-v1/lora_step_0035000.npz b/previous_single_task/LiquidMeasuring-v1/lora_step_0035000.npz new file mode 100644 index 0000000000000000000000000000000000000000..103cc67746875593f36a90efa81230b697457bb1 --- /dev/null +++ b/previous_single_task/LiquidMeasuring-v1/lora_step_0035000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:6e0a9a12b8f5882b391f082b0640baae09dcab280030c3e437804a56f1fa7a8f +size 1867844866 diff --git a/previous_single_task/LiquidMeasuring-v1/metadata.json b/previous_single_task/LiquidMeasuring-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..fb114a91bf73a351c671eb005ca5052a51d8110f --- /dev/null +++ b/previous_single_task/LiquidMeasuring-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "LiquidMeasuring-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0035000.npz", + "checkpoint_sha256": "6e0a9a12b8f5882b391f082b0640baae09dcab280030c3e437804a56f1fa7a8f", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 35000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "547518fad2cc384726b451ac15bf45b72d2e589f1c6eff264781390cabd0fd95", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_LiquidMeasuring-v1", + "source_path": "maniladder/envs/deformable/single_arm_gripper/level_3/liquid_measure.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/LiquidMeasuring-v1/norm_stats.json b/previous_single_task/LiquidMeasuring-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..4480fd356c395a7c8748afe75721b67a02159331 --- /dev/null +++ b/previous_single_task/LiquidMeasuring-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.007872454822063446, + 0.010676166042685509, + -0.0011557338293641806, + 0.006929934024810791, + -0.00008146947948262095, + -0.015435222536325455, + -0.7002655863761902, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.029309742152690887, + 0.04685655236244202, + 0.053979698568582535, + 0.03733079507946968, + 0.007049132138490677, + 0.036248669028282166, + 0.7138824462890625, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.0500043091326952, + -0.10000000149011612, + -0.10000000149011612, + -0.11999999731779099, + -0.026340992639958857, + -0.10951577659402974, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0999614795818925, + 0.09996000148952008, + 0.09996000148952008, + 0.07918759257793428, + 0.004243879737332465, + 0.0008635953575372723, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/LiquidMeasuring-v1/source_audit.json b/previous_single_task/LiquidMeasuring-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..56f568ce5f54189c388781ed56aafa3d213bba05 --- /dev/null +++ b/previous_single_task/LiquidMeasuring-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "LiquidMeasuring-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.4", + "peak_step": "25000", + "final": "0.2", + "start": "2026-09-04 06:14:29", + "eval_mtime": "2026-09-08 10:17:09.036832", + "source_mtime": "2026-08-28 05:07:31.760078", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/single_arm_gripper/level_3/liquid_measure.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_LiquidMeasuring-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_LiquidMeasuring-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_LiquidMeasuring-v1/lora_params/lora_step_0025000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/MazeBlockPush-v1/eval_result.json b/previous_single_task/MazeBlockPush-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..c7621b1f371d8bca5228bb5dde0fbe4d772778f9 --- /dev/null +++ b/previous_single_task/MazeBlockPush-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "MazeBlockPush-v1", + "level": "level_3", + "suite": "rigid_body/single_arm_gripper", + "step": 30000, + "episodes": 20, + "successes": 13, + "success_rate": 0.65, + "return_mean": -240.52585363611578, + "return_mean_finite": true, + "episode_seed_base": null, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "historical_online_eval20", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/MazeBlockPush-v1/evaluation_history.csv b/previous_single_task/MazeBlockPush-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..cebf7e7d4da95a87b9087a31848ff62fed2645ec --- /dev/null +++ b/previous_single_task/MazeBlockPush-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,20,0,0.0,-561.5012655600906,True +10000,20,0,0.0,-550.8270059488714,True +15000,20,5,0.25,-378.9317231580615,True +20000,20,7,0.35,-308.994995149225,True +25000,20,4,0.2,-479.53692783117293,True +30000,20,13,0.65,-240.52585363611578,True +35000,20,8,0.4,-340.74385006651283,True +40000,20,13,0.65,-267.8915696695447,True diff --git a/previous_single_task/MazeBlockPush-v1/metadata.json b/previous_single_task/MazeBlockPush-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..efdae58ae9f236a659ae131c2bb22fc2d062eac7 --- /dev/null +++ b/previous_single_task/MazeBlockPush-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "MazeBlockPush-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0030000.npz", + "checkpoint_sha256": "e7396df9ba9f96b5083f9ca0b1e128b7ce0bc88b8eb5319987eb033f60269602", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 30000, + "eval_episodes": 20, + "eval_seed_base": null, + "norm_stats_sha256": "cfbd7f2fdfd3785dadcf3d8b0ea8e169c333bc08b98247901cff455e724fc638", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_MazeBlockPush-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_3/maze_ball_push.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/MazeBlockPush-v1/norm_stats.json b/previous_single_task/MazeBlockPush-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..b4518d9f03c0c4789936788e07790a10130b9346 --- /dev/null +++ b/previous_single_task/MazeBlockPush-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0074974787421524525, + 0.007390511687844992, + -0.010700209997594357, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.03697575256228447, + 0.04817231371998787, + 0.07186765968799591, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/MazeBlockPush-v1/source_audit.json b/previous_single_task/MazeBlockPush-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..8cb70a5317a5c12c4ae7f0b28ce3e1e1e64fc126 --- /dev/null +++ b/previous_single_task/MazeBlockPush-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "MazeBlockPush-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.65", + "peak_step": "30000", + "final": "0.65", + "start": "2026-09-09 14:42:19", + "eval_mtime": "2026-09-12 19:32:24.571005", + "source_mtime": "2026-09-09 13:32:59.600364", + "reason": "9\u67089\u65e513:32\u6539\u4e3a\u65e0seed\u72ec\u7acb\u968f\u673a\u8ff7\u5bab\uff1b14:42\u91cd\u65b0\u542f\u52a8\u5355\u4efb\u52a1\u8bad\u7ec3\uff0c9\u670812\u65e5\u5b8c\u6210\uff1b\u6700\u65b0\u5cf0\u503c0.65@30k\u66ff\u4ee3\u65e7\u6c47\u603b0.60", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_3/maze_ball_push.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_MazeBlockPush-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_MazeBlockPush-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_MazeBlockPush-v1/lora_params/lora_step_0030000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/MobilePickPlace-v1/eval_result.json b/previous_single_task/MobilePickPlace-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..95b89c3dd625909ad4fd212ee1838708a6665bdb --- /dev/null +++ b/previous_single_task/MobilePickPlace-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "MobilePickPlace-v1", + "level": "level_3", + "suite": "rigid_body/single_arm_dex", + "step": 15000, + "episodes": 50, + "successes": 4, + "success_rate": 0.08, + "return_mean": null, + "return_mean_finite": false, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/MobilePickPlace-v1/evaluation_history.csv b/previous_single_task/MobilePickPlace-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..efe45088b2f14f18a4e56f767f43ed55a0ef6ed6 --- /dev/null +++ b/previous_single_task/MobilePickPlace-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,2,0.04,,False +10000,50,0,0.0,,False +15000,50,4,0.08,,False +20000,50,3,0.06,,False +25000,50,1,0.02,,False +30000,50,1,0.02,,False +35000,50,2,0.04,169.25979777826228,True +40000,50,2,0.04,,False diff --git a/previous_single_task/MobilePickPlace-v1/lora_step_0015000.npz b/previous_single_task/MobilePickPlace-v1/lora_step_0015000.npz new file mode 100644 index 0000000000000000000000000000000000000000..8ef8ecab430924e1905fde086f25cfd07141d6be --- /dev/null +++ b/previous_single_task/MobilePickPlace-v1/lora_step_0015000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:a5f73d09c256f142d64b4edf03062bc328f6a3d0c13c85a0949a04d875cd64cb +size 1867844866 diff --git a/previous_single_task/MobilePickPlace-v1/metadata.json b/previous_single_task/MobilePickPlace-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..3dc42fef1ff676cc1fe53b7306fadb85ec6a2014 --- /dev/null +++ b/previous_single_task/MobilePickPlace-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "MobilePickPlace-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0015000.npz", + "checkpoint_sha256": "a5f73d09c256f142d64b4edf03062bc328f6a3d0c13c85a0949a04d875cd64cb", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 15000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "3e0618e0441de4799148f7725eba884b4732ab010ce526c65c93932462a82573", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_MobilePickPlace-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_3/mobile_pick_place.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/MobilePickPlace-v1/norm_stats.json b/previous_single_task/MobilePickPlace-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..d0918dd39c6e90f88bf78831b049951bc97b70dc --- /dev/null +++ b/previous_single_task/MobilePickPlace-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.47651559114456177, + -0.06437011063098907, + -1.3807376623153687, + 0.0012244728859513998, + -0.009673789143562317, + -0.007055480033159256, + -7.021400051598903e-6, + 0.000030506054827128537, + 0.00017484667478129268, + 0.0008526229648850858, + 0.015098723582923412, + 0.019113805145025253, + 0.01269460003823042, + 0.0002626645436976105, + 0.015509091317653656, + 0.01827762834727764, + 0.011913688853383064, + 0.00026014173636212945, + 0.015171640552580357, + 0.017365964129567146, + 0.011108958162367344, + 0.03215735778212547, + -0.003501489059999585, + 0.1009136438369751, + 0.05471954494714737, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.6589765548706055, + 0.26602309942245483, + 1.376226782798767, + 0.08463605493307114, + 0.056461334228515625, + 0.13501062989234924, + 0.0012551594991236925, + 0.009104038588702679, + 0.0025801428128033876, + 0.001682784641161561, + 0.021559204906225204, + 0.12794403731822968, + 0.1276775300502777, + 0.0011724610812962055, + 0.02218150533735752, + 0.12897618114948273, + 0.1285741627216339, + 0.0010293764062225819, + 0.02196413464844227, + 0.12985600531101227, + 0.12928476929664612, + 0.17614196240901947, + 0.02790611982345581, + 0.18179430067539215, + 0.15063878893852234, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.9784359931945801, + -0.8990269899368286, + -3.142848425102234, + -0.30000001192092896, + -0.30000001192092896, + -0.30000001192092896, + -0.0016890012165531529, + -0.0215438172236085, + -0.0003178407512605258, + -0.001553434245288372, + -0.00012505856284406036, + -0.32849549357891084, + -0.3496271916389465, + -0.0015665040811523796, + -0.00011037648477358743, + -0.33917470484972, + -0.35896904268264773, + -0.0012814680818468331, + -0.0001312803360633552, + -0.3459683064341545, + -0.36432598774433134, + -0.5178465687036514, + -0.09322490980178118, + -0.1581073681473732, + -0.250015711069107, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.9993293519735336, + 0.08912378716021785, + 1.5695431898117063, + 0.29988001191616065, + 0.1288853346347809, + 0.29988001191616065, + 0.0016226808357983832, + 0.023713074076175678, + 0.003300689116120338, + 0.005226726904511452, + 0.07591155585568632, + 0.4080476223468781, + 0.3936389277458191, + 0.0038117084970697764, + 0.07917155532240869, + 0.40786184475421905, + 0.3937428803682327, + 0.0037767172262072567, + 0.08171145514931995, + 0.40781300163865086, + 0.3937353855609894, + 0.6266077923178675, + 0.07807590529471636, + 0.45529925907850277, + 0.4391681716918945, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/MobilePickPlace-v1/source_audit.json b/previous_single_task/MobilePickPlace-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..48184321a27c09b457c7f3eeedc76bcfc5c86cbe --- /dev/null +++ b/previous_single_task/MobilePickPlace-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "MobilePickPlace-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.2", + "peak_step": "25000", + "final": "0.05", + "start": "2026-08-22 06:21:14", + "eval_mtime": "2026-08-25 13:22:33.374291", + "source_mtime": "2026-08-10 09:04:08.001709", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_3/mobile_pick_place.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_MobilePickPlace-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_MobilePickPlace-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_MobilePickPlace-v1/lora_params/lora_step_0025000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/OpenDoor-v1/eval_result.json b/previous_single_task/OpenDoor-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..b2d9e24233886862efc8c060fae1b58c352520ed --- /dev/null +++ b/previous_single_task/OpenDoor-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "OpenDoor-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 15, + "success_rate": 0.3, + "return_mean": 81.66296870330603, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/OpenDoor-v1/evaluation_history.csv b/previous_single_task/OpenDoor-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..2366e1a34125f37217e2d9cd2521e68542a6b2ea --- /dev/null +++ b/previous_single_task/OpenDoor-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,15,0.3,81.66296870330603,True +10000,50,12,0.24,93.23127404245614,True +15000,50,12,0.24,47.44674762370511,True +20000,50,10,0.2,80.22020105135914,True +25000,50,10,0.2,69.365043020613,True +30000,50,13,0.26,60.57508714767713,True +35000,50,11,0.22,52.8616152550091,True +40000,50,9,0.18,61.84445682378205,True diff --git a/previous_single_task/OpenDoor-v1/lora_step_0005000.npz b/previous_single_task/OpenDoor-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..7df4e2a1219d9c0a1efc6b5ffc225aa226920a22 --- /dev/null +++ b/previous_single_task/OpenDoor-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9becc4d5effe10eca7baef2875443f2e5ab55c21ec18fbc863537e30280973f7 +size 1867844866 diff --git a/previous_single_task/OpenDoor-v1/metadata.json b/previous_single_task/OpenDoor-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..c3921c53702e6201b75e84674514750b6869d558 --- /dev/null +++ b/previous_single_task/OpenDoor-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "OpenDoor-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "9becc4d5effe10eca7baef2875443f2e5ab55c21ec18fbc863537e30280973f7", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "3e16febbd98c953cb6bae4c18add409bec5b3cb3b8cdbafcdf2ada2678b4050f", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_OpenDoor-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_1/open_door.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/OpenDoor-v1/norm_stats.json b/previous_single_task/OpenDoor-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..2b44e4138d2f9a17b68151b8f6ad590e9bd77975 --- /dev/null +++ b/previous_single_task/OpenDoor-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0009262226521968842, + -0.008006278425455093, + -0.012124485336244106, + 0.0, + 0.0, + -0.03420540317893028, + 0.16249997913837433, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.04506584256887436, + 0.04418841749429703, + 0.040248509496450424, + 0.0, + 0.0, + 0.029703568667173386, + 0.9867085814476013, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + -0.05999999865889549, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + -0.000011999999731782118, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/OpenDoor-v1/source_audit.json b/previous_single_task/OpenDoor-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..99ee8774191c341218cf88bd9ebe11bbc33c79b7 --- /dev/null +++ b/previous_single_task/OpenDoor-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "OpenDoor-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.75", + "peak_step": "20000", + "final": "0.6", + "start": "2026-08-21 13:37:04", + "eval_mtime": "2026-08-25 06:42:15.060887", + "source_mtime": "2026-05-27 05:38:46.450865", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_1/open_door.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_OpenDoor-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_OpenDoor-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_OpenDoor-v1/lora_params/lora_step_0020000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/PickPlacePen-v1/eval_result.json b/previous_single_task/PickPlacePen-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..c3a58f5ca4d0ed379c01eb3a616e52fd5a6e36ca --- /dev/null +++ b/previous_single_task/PickPlacePen-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "PickPlacePen-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_gripper", + "step": 20000, + "episodes": 50, + "successes": 18, + "success_rate": 0.36, + "return_mean": 95.0536512184143, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/PickPlacePen-v1/evaluation_history.csv b/previous_single_task/PickPlacePen-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..cca58503f3942be94d8fda6d4af4d7dc189e7d6c --- /dev/null +++ b/previous_single_task/PickPlacePen-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,10,0.2,58.34850861787796,True +10000,50,13,0.26,102.91128782510758,True +15000,50,17,0.34,109.48091080307961,True +20000,50,18,0.36,95.0536512184143,True +25000,50,13,0.26,111.45588166356087,True +30000,50,11,0.22,117.25063677191734,True +35000,50,13,0.26,117.22585187792778,True +40000,50,16,0.32,112.43077191233635,True diff --git a/previous_single_task/PickPlacePen-v1/lora_step_0020000.npz b/previous_single_task/PickPlacePen-v1/lora_step_0020000.npz new file mode 100644 index 0000000000000000000000000000000000000000..f9b62382e3e9e56d1c53dc94a1bcb38702221cd2 --- /dev/null +++ b/previous_single_task/PickPlacePen-v1/lora_step_0020000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:8969ee8461c5a3ce36cb0a205474c1214990589054aa85ce2f3b1f707e1ab16f +size 1867844866 diff --git a/previous_single_task/PickPlacePen-v1/metadata.json b/previous_single_task/PickPlacePen-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..857e469ecab5d5407fc83a90cf4e0ea16f2dbf07 --- /dev/null +++ b/previous_single_task/PickPlacePen-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "PickPlacePen-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0020000.npz", + "checkpoint_sha256": "8969ee8461c5a3ce36cb0a205474c1214990589054aa85ce2f3b1f707e1ab16f", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 20000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "b3d3c93215fedaf4534454c63056d20ecad7ea303b3b649dd19b14329e1947f5", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_PickPlacePen-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_1/pick_place_pen.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/PickPlacePen-v1/norm_stats.json b/previous_single_task/PickPlacePen-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..3401584f25bbca977fcac7260d0edb24b4eca8dc --- /dev/null +++ b/previous_single_task/PickPlacePen-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.009110755287110806, + 0.0016610411694273353, + -0.0016215266659855843, + 0.0, + -0.008067211136221886, + -0.00007091020961524919, + -0.04585597291588783, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.027197057381272316, + 0.04785013943910599, + 0.07118619978427887, + 0.0, + 0.06101062893867493, + 0.003203312400728464, + 0.9989480376243591, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.00004000000059603548, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + -0.20000000298023224, + -0.006195582278445355, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.19992000297904017, + 0.004413731339946408, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/PickPlacePen-v1/source_audit.json b/previous_single_task/PickPlacePen-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..c74443c4cd40bd20783bbe3593135294d3029ef2 --- /dev/null +++ b/previous_single_task/PickPlacePen-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "PickPlacePen-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.55", + "peak_step": "20000", + "final": "0.45", + "start": "2026-08-22 06:11:20", + "eval_mtime": "2026-08-26 04:25:14.003450", + "source_mtime": "2026-02-17 04:31:14.274228", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_1/pick_place_pen.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_PickPlacePen-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_PickPlacePen-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_PickPlacePen-v1/lora_params/lora_step_0020000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/PingPongBallTransport-v1/eval_result.json b/previous_single_task/PingPongBallTransport-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..10c394aac345aa864d71a96097dbd2caf0d1c35a --- /dev/null +++ b/previous_single_task/PingPongBallTransport-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "PingPongBallTransport-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_gripper", + "step": 40000, + "episodes": 50, + "successes": 8, + "success_rate": 0.16, + "return_mean": 1567.3499554800987, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 7, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/PingPongBallTransport-v1/evaluation_history.csv b/previous_single_task/PingPongBallTransport-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..ef724620bcce3eee63baec521cff7cf04e53c0a0 --- /dev/null +++ b/previous_single_task/PingPongBallTransport-v1/evaluation_history.csv @@ -0,0 +1,8 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,280.424271569252,True +15000,50,1,0.02,481.7520525765419,True +20000,50,3,0.06,1191.8369809103012,True +25000,50,5,0.1,1601.3661450529098,True +30000,50,0,0.0,828.809775466919,True +35000,50,4,0.08,1499.5457893419266,True +40000,50,8,0.16,1567.3499554800987,True diff --git a/previous_single_task/PingPongBallTransport-v1/lora_step_0040000.npz b/previous_single_task/PingPongBallTransport-v1/lora_step_0040000.npz new file mode 100644 index 0000000000000000000000000000000000000000..2472f0488468fbfa0d7ae2ee8d7bfb9ab7dac5b6 --- /dev/null +++ b/previous_single_task/PingPongBallTransport-v1/lora_step_0040000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:4f62cb281fe6126427b03b9ee60d949ffd7cd6ce38d4b18d7b8203b831d0261a +size 1867844866 diff --git a/previous_single_task/PingPongBallTransport-v1/metadata.json b/previous_single_task/PingPongBallTransport-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..6f6f6fe035b93773732527c0f8bf5956332b1d31 --- /dev/null +++ b/previous_single_task/PingPongBallTransport-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "PingPongBallTransport-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0040000.npz", + "checkpoint_sha256": "4f62cb281fe6126427b03b9ee60d949ffd7cd6ce38d4b18d7b8203b831d0261a", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 40000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "980764473bad4e83ed65cd1bbb4394fa90d0d5d30fe8656de5d0b78ad6302aab", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_PingPongBallTransport-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_4/pingpong_ball_transport.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/PingPongBallTransport-v1/norm_stats.json b/previous_single_task/PingPongBallTransport-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..e8152f13c3ea96d5e3f743ade8e47f6661ee6866 --- /dev/null +++ b/previous_single_task/PingPongBallTransport-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.0024766961578279734, + -0.00005698251334251836, + -0.004120966885238886, + 0.0, + -0.0011382593074813485, + 0.0, + 0.17391298711299896, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.06139069423079491, + 0.04276640713214874, + 0.04075935110449791, + 0.0, + 0.03586148843169212, + 0.0, + 0.9376965165138245, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.11999999731779099, + -0.11999999731779099, + -0.11999999731779099, + 0.0, + -0.12566371262073517, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.11995199731886386, + 0.11995199731886386, + 0.059963998659700146, + 0.0, + 0.12561344713568684, + 0.0, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/PingPongBallTransport-v1/source_audit.json b/previous_single_task/PingPongBallTransport-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..e1a8ae7e179616ec41f5349b76dea1272ec9844f --- /dev/null +++ b/previous_single_task/PingPongBallTransport-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "PingPongBallTransport-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.1", + "peak_step": "25000", + "final": "0.05", + "start": "2026-08-21 14:14:37", + "eval_mtime": "2026-08-25 02:48:17.956538", + "source_mtime": "2026-05-26 12:28:16.489932", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_4/pingpong_ball_transport.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_PingPongBallTransport-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_PingPongBallTransport-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_PingPongBallTransport-v1/lora_params/lora_step_0025000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/PivotBook-v1/eval_result.json b/previous_single_task/PivotBook-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..a6cadd350c8918b112f4f9a69c3c4b16ccfbfd64 --- /dev/null +++ b/previous_single_task/PivotBook-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "PivotBook-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_gripper", + "step": 15000, + "episodes": 50, + "successes": 19, + "success_rate": 0.38, + "return_mean": 216.95523127913475, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/PivotBook-v1/evaluation_history.csv b/previous_single_task/PivotBook-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..b37deccae7b513374b04a520a49098bb4bab5b37 --- /dev/null +++ b/previous_single_task/PivotBook-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,8,0.16,236.53256025791168,True +10000,50,10,0.2,143.90750781297683,True +15000,50,19,0.38,216.95523127913475,True +20000,50,13,0.26,193.05574815273286,True +25000,50,14,0.28,231.08132846355437,True +30000,50,14,0.28,307.9155409657955,True +35000,50,14,0.28,172.2963804090023,True +40000,50,16,0.32,197.24141772389413,True diff --git a/previous_single_task/PivotBook-v1/lora_step_0015000.npz b/previous_single_task/PivotBook-v1/lora_step_0015000.npz new file mode 100644 index 0000000000000000000000000000000000000000..eadfe506d585353938c6092b2b02ef015b6fb1d3 --- /dev/null +++ b/previous_single_task/PivotBook-v1/lora_step_0015000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1f419832345ac22ea2f4a05686c32c7670f62d2704c5b5025e19414cfa51b047 +size 1867844866 diff --git a/previous_single_task/PivotBook-v1/metadata.json b/previous_single_task/PivotBook-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..63e7bf8aed94589dc327ef3f6cab6a9cabe3fce8 --- /dev/null +++ b/previous_single_task/PivotBook-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "PivotBook-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0015000.npz", + "checkpoint_sha256": "1f419832345ac22ea2f4a05686c32c7670f62d2704c5b5025e19414cfa51b047", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 15000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "44874fe8a32912daff526f7dcb9b1f573488a724ea8a0a640c1a1750d78a12bc", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_PivotBook-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_2/pivot_book.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/PivotBook-v1/norm_stats.json b/previous_single_task/PivotBook-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..82572556da603bfc8c815b1818acbad8c259cc25 --- /dev/null +++ b/previous_single_task/PivotBook-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.0013994508190080523, + 0.0024921842850744724, + -0.01019515935331583, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0598938874900341, + 0.07706094533205032, + 0.05484481900930405, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10001888144612311, + -0.20000000298023224, + -0.10001874425709248, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09995357185155151, + 0.19992000297904017, + 0.1367403894394636, + 0.0, + 0.0, + 0.0, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/PivotBook-v1/source_audit.json b/previous_single_task/PivotBook-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..f21c75c9fb60eb05a8eb1439e35c90ddc715dafb --- /dev/null +++ b/previous_single_task/PivotBook-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "PivotBook-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.4", + "peak_step": "10000", + "final": "0.1", + "start": "2026-08-22 05:40:30", + "eval_mtime": "2026-08-25 04:02:42.472404", + "source_mtime": "2026-07-02 12:27:13.105724", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_2/pivot_book.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_PivotBook-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_PivotBook-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_PivotBook-v1/lora_params/lora_step_0010000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/PushSoccer-v1/eval_result.json b/previous_single_task/PushSoccer-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..694d50ae8263635c83266291294171d157f42672 --- /dev/null +++ b/previous_single_task/PushSoccer-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "PushSoccer-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_gripper", + "step": 40000, + "episodes": 50, + "successes": 40, + "success_rate": 0.8, + "return_mean": 55.96963231444359, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/PushSoccer-v1/evaluation_history.csv b/previous_single_task/PushSoccer-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..3cb4971e315b5a808d7b1e6e96dd86fdf376a747 --- /dev/null +++ b/previous_single_task/PushSoccer-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,32,0.64,63.92137126088142,True +10000,50,33,0.66,65.62150128126144,True +15000,50,36,0.72,61.28525426745415,True +20000,50,39,0.78,53.52894441485405,True +25000,50,38,0.76,55.87276070356369,True +30000,50,38,0.76,60.96458682656288,True +35000,50,38,0.76,56.6934594810009,True +40000,50,40,0.8,55.96963231444359,True diff --git a/previous_single_task/PushSoccer-v1/lora_step_0040000.npz b/previous_single_task/PushSoccer-v1/lora_step_0040000.npz new file mode 100644 index 0000000000000000000000000000000000000000..79c9c2a028bfb54e501fe62c6b3ef19ec933d503 --- /dev/null +++ b/previous_single_task/PushSoccer-v1/lora_step_0040000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:b495183247ad15943fb3abd608c423928d8ceba2859efef3380fcc4818297a54 +size 1867844866 diff --git a/previous_single_task/PushSoccer-v1/metadata.json b/previous_single_task/PushSoccer-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..e642c84d5a781d879c3096eb2b5af491b21333ce --- /dev/null +++ b/previous_single_task/PushSoccer-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "PushSoccer-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0040000.npz", + "checkpoint_sha256": "b495183247ad15943fb3abd608c423928d8ceba2859efef3380fcc4818297a54", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 40000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "9c2a5b81ee826e2c453938184b2c1bf9dcd34f6e2e3004cb466d3f61b79ec819", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_PushSoccer-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_1/push_soccer.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/PushSoccer-v1/norm_stats.json b/previous_single_task/PushSoccer-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..18471918f8adff95995e1c8ecc5a5de588dee414 --- /dev/null +++ b/previous_single_task/PushSoccer-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.027352437376976013, + -0.0008789423736743629, + -0.027176689356565475, + 0.0, + 0.0, + -0.001510332920588553, + 1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.047818586230278015, + 0.06765623390674591, + 0.05444752052426338, + 0.0, + 0.0, + 0.042741261422634125, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + -0.1599999964237213, + 1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.15993599642515183, + 1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/PushSoccer-v1/source_audit.json b/previous_single_task/PushSoccer-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..49d863d97ed31443bc2b76993a7534ed6217e3ee --- /dev/null +++ b/previous_single_task/PushSoccer-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "PushSoccer-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.85", + "peak_step": "25000", + "final": "0.7", + "start": "2026-08-21 13:37:06", + "eval_mtime": "2026-08-24 20:37:39.205981", + "source_mtime": "2026-05-25 13:23:03.864694", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_1/push_soccer.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_PushSoccer-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_PushSoccer-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_PushSoccer-v1/lora_params/lora_step_0025000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/ReorientPen-v1/eval_result.json b/previous_single_task/ReorientPen-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..ecd621fdb660f818a2762d7138592e7881a1f015 --- /dev/null +++ b/previous_single_task/ReorientPen-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "ReorientPen-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_dex", + "step": 25000, + "episodes": 50, + "successes": 19, + "success_rate": 0.38, + "return_mean": 104.46506033671089, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/ReorientPen-v1/evaluation_history.csv b/previous_single_task/ReorientPen-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..d1c0996c6caf5b55c8d03a958c9f289438e00d37 --- /dev/null +++ b/previous_single_task/ReorientPen-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,13,0.26,111.70928847831674,True +10000,50,16,0.32,109.54125524462201,True +15000,50,18,0.36,108.74469565531238,True +20000,50,14,0.28,110.39952756698709,True +25000,50,19,0.38,104.46506033671089,True +30000,50,14,0.28,106.68186323479168,True +35000,50,15,0.3,104.7015729028359,True +40000,50,14,0.28,105.2235934635438,True diff --git a/previous_single_task/ReorientPen-v1/lora_step_0025000.npz b/previous_single_task/ReorientPen-v1/lora_step_0025000.npz new file mode 100644 index 0000000000000000000000000000000000000000..26c085f9fee24c6cfb771b2430ca4f95ad752bda --- /dev/null +++ b/previous_single_task/ReorientPen-v1/lora_step_0025000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:070504c2d43469bc87fd254b20ab68c2c4b646c9acf7fc066a6a5ad0e5bde174 +size 1867844866 diff --git a/previous_single_task/ReorientPen-v1/metadata.json b/previous_single_task/ReorientPen-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..3ffd0ea48c2e93c9e44f5755121ab926d2bbb157 --- /dev/null +++ b/previous_single_task/ReorientPen-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "ReorientPen-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0025000.npz", + "checkpoint_sha256": "070504c2d43469bc87fd254b20ab68c2c4b646c9acf7fc066a6a5ad0e5bde174", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 25000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "b09166da1b1b17fbf02410147ae519981795c57e0ac7f163f6f65af50faf9d4e", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_ReorientPen-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_2/reorient_pen.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/ReorientPen-v1/norm_stats.json b/previous_single_task/ReorientPen-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..35147d182af776e6473323af036161a3edd07603 --- /dev/null +++ b/previous_single_task/ReorientPen-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.00009943910117726773, + 0.01957705430686474, + 0.0007601366960443556, + -0.00757370050996542, + -0.01025445107370615, + -0.013631586916744709, + -0.0184859037399292, + 0.08119163662195206, + 0.12791676819324493, + 0.0, + 0.011710486374795437, + 0.08013193309307098, + 0.12287013977766037, + 0.0, + -0.05582267418503761, + -0.032233577221632004, + 0.12920941412448883, + 0.0, + 0.05524757504463196, + -0.0002489863836672157, + 0.018308265134692192, + 0.00016653469356242567, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.031094161793589592, + 0.04649551957845688, + 0.07837500423192978, + 0.07446905970573425, + 0.07291419804096222, + 0.09828197956085205, + 0.1722608059644699, + 0.34662923216819763, + 0.575851321220398, + 0.0, + 0.5210866332054138, + 0.31253498792648315, + 0.6383266448974609, + 0.0, + 0.2561972141265869, + 0.17668430507183075, + 0.6399807929992676, + 0.0, + 0.24889309704303741, + 0.02176564931869507, + 0.11967980116605759, + 0.009571390226483345, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.0749876397550106, + -0.07269617246985435, + -0.16706641137003897, + -0.2769007217854262, + -0.28010785295665264, + -0.49479392715990544, + -0.8044946324825286, + -0.2914368414878843, + -0.4713946580886841, + 0.0, + -1.7802009860038757, + -0.17606749010086054, + -0.43351973438262936, + 0.0, + -1.5087166738510132, + -1.025383533143997, + -0.42584553480148335, + 0.0, + -0.1873679590225219, + -0.04798581225275994, + -0.272711443901062, + -0.0314886729478836, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09116274925470352, + 0.14926805713772776, + 0.21902772520780567, + 0.13982962841689583, + 0.1397971479386091, + 0.1397297605246306, + 0.364541160941124, + 1.9615460634231567, + 3.0512085437774656, + 0.0, + 2.871056961345672, + 1.8618134329318998, + 3.7481980419158933, + 0.0, + 0.13233652353286773, + 0.24045861775875088, + 3.7880718541145324, + 0.0, + 1.166383566379547, + 0.054605780559778216, + 0.4394293308258057, + 0.021614997494220745, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/ReorientPen-v1/source_audit.json b/previous_single_task/ReorientPen-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..c4c9fb549d7326bed4c3d7dfb87157a6852891c2 --- /dev/null +++ b/previous_single_task/ReorientPen-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "ReorientPen-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.45", + "peak_step": "30000", + "final": "0.4", + "start": "2026-08-21 13:37:11", + "eval_mtime": "2026-08-25 06:46:52.497177", + "source_mtime": "2026-01-13 12:49:03.651940", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_2/reorient_pen.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_ReorientPen-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_ReorientPen-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_ReorientPen-v1/lora_params/lora_step_0030000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/RingTossing-v1/eval_result.json b/previous_single_task/RingTossing-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..5544f766d0bd82e7798cab5c307ae12a86e1d2e4 --- /dev/null +++ b/previous_single_task/RingTossing-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "RingTossing-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": -2867.2832674495876, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/RingTossing-v1/evaluation_history.csv b/previous_single_task/RingTossing-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..ebff823a4acac0f25bf5e2bd6ccf1eb576b5f71a --- /dev/null +++ b/previous_single_task/RingTossing-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,-2867.2832674495876,True +10000,50,0,0.0,-3205.7201180787383,True +15000,50,0,0.0,-3219.2910319840908,True +20000,50,0,0.0,-3174.6684483157096,True +25000,50,0,0.0,-3033.330185411349,True +30000,50,0,0.0,-3175.8641105918587,True +35000,50,0,0.0,-3372.0425525617597,True +40000,50,0,0.0,-3147.3354511817543,True diff --git a/previous_single_task/RingTossing-v1/lora_step_0005000.npz b/previous_single_task/RingTossing-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..45c00cc6ffa0ffeb8c9216e3276046ba492406f4 --- /dev/null +++ b/previous_single_task/RingTossing-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:5b01cd6223af4a4e8cd3a94acd9d6a4c50782845f804a385b4b5719794ede293 +size 1867844866 diff --git a/previous_single_task/RingTossing-v1/metadata.json b/previous_single_task/RingTossing-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..6ed7bc923139c4f64adbb4b229564c272f225c74 --- /dev/null +++ b/previous_single_task/RingTossing-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "RingTossing-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "5b01cd6223af4a4e8cd3a94acd9d6a4c50782845f804a385b4b5719794ede293", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "e3b2580d01b736bcd4f3a908d67d73d568cc5c9eaf4113e11cf520e56fac5812", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_RingTossing-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_4/ring_tossing.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/RingTossing-v1/norm_stats.json b/previous_single_task/RingTossing-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..750cbac27bbf2bad299f124672e92eb9036c5691 --- /dev/null +++ b/previous_single_task/RingTossing-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.027247652411460876, + -0.009958493523299694, + -0.0043901801109313965, + 0.0016672596102580428, + 0.0020240547601133585, + 0.005483911372721195, + 0.21946027874946594, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.10938502848148346, + 0.08641652017831802, + 0.13654395937919617, + 0.0316697359085083, + 0.03301055356860161, + 0.032600194215774536, + 0.9756214618682861, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.11999999731779099, + -0.2639279917597771, + -0.12011999624967573, + -0.07999999821186066, + -0.07999999821186066, + -0.07999999821186066, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.4198919868901372, + 0.17707199447154998, + 0.41983198689222334, + 0.07996799821257591, + 0.07996799821257591, + 0.07996799821257591, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/RingTossing-v1/source_audit.json b/previous_single_task/RingTossing-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..0ec24d8814affd58cbddc31ba08e735392275db6 --- /dev/null +++ b/previous_single_task/RingTossing-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "RingTossing-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 06:11:23", + "eval_mtime": "2026-08-26 12:21:04.717862", + "source_mtime": "2026-07-31 06:10:31.167273", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_4/ring_tossing.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_RingTossing-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_RingTossing-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_RingTossing-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/RodThroughHoles-v1/eval_result.json b/previous_single_task/RodThroughHoles-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..fbc5bca71ed9b621969c21b411bca71081560a2b --- /dev/null +++ b/previous_single_task/RodThroughHoles-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "RodThroughHoles-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_dex", + "step": 5000, + "episodes": 50, + "successes": 1, + "success_rate": 0.02, + "return_mean": 678.4349171388521, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/RodThroughHoles-v1/evaluation_history.csv b/previous_single_task/RodThroughHoles-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..fd5241260ccfd61ef4b2c2461aa54ea531b02cc0 --- /dev/null +++ b/previous_single_task/RodThroughHoles-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,1,0.02,678.4349171388521,True +10000,50,1,0.02,724.497755086273,True +15000,50,0,0.0,560.2698704089597,True +20000,50,0,0.0,656.1754601595178,True +25000,50,0,0.0,648.2634786164016,True +30000,50,0,0.0,,False +35000,50,0,0.0,786.6771039431542,True +40000,50,0,0.0,695.7203879581019,True diff --git a/previous_single_task/RodThroughHoles-v1/lora_step_0005000.npz b/previous_single_task/RodThroughHoles-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..879ae77d944cabdac79ae71616efece38d5f6438 --- /dev/null +++ b/previous_single_task/RodThroughHoles-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:21e9f0483831cd65a5e75e8415b9b1c496e7ed1714b3b85a1f685184abf4a6ca +size 1867844866 diff --git a/previous_single_task/RodThroughHoles-v1/metadata.json b/previous_single_task/RodThroughHoles-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..745d0af650d3aceffc94cddd28ee08fbd02056f7 --- /dev/null +++ b/previous_single_task/RodThroughHoles-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "RodThroughHoles-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "21e9f0483831cd65a5e75e8415b9b1c496e7ed1714b3b85a1f685184abf4a6ca", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "58820fcd228391872bccd5bdd5c109b0927312a802a7652392557858cf994785", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_RodThroughHoles-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_4/rod_through_holes.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/RodThroughHoles-v1/norm_stats.json b/previous_single_task/RodThroughHoles-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..3d4fd8ea763ece9b1b996c01ee1b512ab877d005 --- /dev/null +++ b/previous_single_task/RodThroughHoles-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.02416636236011982, + 0.00436018593609333, + -0.007039946038275957, + 0.0, + 0.007538671139627695, + -0.011774160899221897, + -0.023348994553089142, + 0.13325899839401245, + 0.1516621708869934, + 0.08319126814603806, + 0.008974677883088589, + 0.10735713690519333, + 0.13229458034038544, + 0.08565616607666016, + 0.011693915352225304, + 0.061180274933576584, + 0.0712614506483078, + 0.05050526559352875, + 0.02593652345240116, + -0.02036762610077858, + 0.2562161684036255, + 0.13450689613819122, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.06211244687438011, + 0.092173732817173, + 0.08225784450769424, + 0.0, + 0.027097569778561592, + 0.03386295586824417, + 0.02346484549343586, + 0.058898039162158966, + 0.04895487427711487, + 0.05057396739721298, + 0.01357338484376669, + 0.05909668281674385, + 0.05037674680352211, + 0.050388459116220474, + 0.015648148953914642, + 0.07795332372188568, + 0.07989702373743057, + 0.06699356436729431, + 0.04542071372270584, + 0.01820502243936062, + 0.05644659698009491, + 0.04398157820105553, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.20000000298023224, + -0.20000000298023224, + -0.20000000298023224, + 0.0, + -0.053520000797510146, + -0.10000000149011612, + -0.12856983644422143, + -0.00001405910552421119, + -2.2326535145111848e-6, + -8.631210413057033e-9, + -0.025687577821314336, + -0.000013809890333504882, + -2.1771688807348255e-6, + -8.635621107089264e-9, + -0.008463395866192876, + -0.0002745714401826264, + -2.1212549654592294e-6, + -8.639754689454548e-9, + -0.009474280485510826, + -0.06726590667366981, + 0.011093683159723878, + 0.011075473423302173, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09424000140428546, + 0.19992000297904017, + 0.19992000297904017, + 0.0, + 0.09996000148952008, + 0.060680000904202475, + 0.00021410873909483152, + 0.39760987365386025, + 0.37111033286586553, + 0.3446039863553277, + 0.03975164502114058, + 0.39761032399582147, + 0.37111039199094104, + 0.3446039863548143, + 0.05438650966826826, + 0.3976010845815763, + 0.37111047749931697, + 0.34460398635433315, + 0.25424712238907815, + 0.011354237315058704, + 0.329546998026967, + 0.3295290162682533, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/RodThroughHoles-v1/source_audit.json b/previous_single_task/RodThroughHoles-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..2654ce09d8ab19160da06cdec9fc230943bea911 --- /dev/null +++ b/previous_single_task/RodThroughHoles-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "RodThroughHoles-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 06:31:07", + "eval_mtime": "2026-08-25 18:41:45.129645", + "source_mtime": "2026-08-09 10:03:02.430739", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_4/rod_through_holes.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_RodThroughHoles-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_RodThroughHoles-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_RodThroughHoles-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/RollingPin-v1/eval_result.json b/previous_single_task/RollingPin-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..fba962f6e3468ad9386a1672520858bc1a67e00d --- /dev/null +++ b/previous_single_task/RollingPin-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "RollingPin-v1", + "level": "level_2", + "suite": "deformable/single_arm_dex", + "step": 25000, + "episodes": 50, + "successes": 48, + "success_rate": 0.96, + "return_mean": 212.23153244376184, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/RollingPin-v1/evaluation_history.csv b/previous_single_task/RollingPin-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..8fa813d98daf9ff6cdedb1a3c27a5c04da75aa8f --- /dev/null +++ b/previous_single_task/RollingPin-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,31,0.62,236.66776097416877,True +10000,50,41,0.82,219.33673781514167,True +15000,50,44,0.88,219.61299183487893,True +20000,50,42,0.84,223.52243975043297,True +25000,50,48,0.96,212.23153244376184,True +30000,50,42,0.84,216.7686476457119,True +35000,50,41,0.82,220.03086215257645,True +40000,50,46,0.92,214.28712272286415,True diff --git a/previous_single_task/RollingPin-v1/lora_step_0025000.npz b/previous_single_task/RollingPin-v1/lora_step_0025000.npz new file mode 100644 index 0000000000000000000000000000000000000000..2eaefbf7c0b0753de7d3f029c14acf415dafd3c1 --- /dev/null +++ b/previous_single_task/RollingPin-v1/lora_step_0025000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:49c64d6d95014fb02dc5fde1dd3adfac52c28264a1fd6a7a7cd8499a06f410ac +size 1867844866 diff --git a/previous_single_task/RollingPin-v1/metadata.json b/previous_single_task/RollingPin-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..5558b6cb1630c653b3f2a94e5c218ddca1dfcf23 --- /dev/null +++ b/previous_single_task/RollingPin-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "RollingPin-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0025000.npz", + "checkpoint_sha256": "49c64d6d95014fb02dc5fde1dd3adfac52c28264a1fd6a7a7cd8499a06f410ac", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 25000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "ed1318239f743540932c10a2d29a730a86cc11477ce95fea503546f6a4825c19", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_RollingPin-v1", + "source_path": "maniladder/envs/deformable/single_arm_dex/level_2/rooling_pin.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/RollingPin-v1/norm_stats.json b/previous_single_task/RollingPin-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..bb6a3dac3c7c6163a078fc88bb948a9f91fa79dd --- /dev/null +++ b/previous_single_task/RollingPin-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.006999306846410036, + 0.014740638434886932, + -0.02603905275464058, + 0.0, + 0.0, + 0.0, + -0.005625956691801548, + 0.11066768318414688, + 0.09807612746953964, + 0.08007576316595078, + -0.00918341800570488, + 0.09084577113389969, + 0.09467066824436188, + 0.0782671868801117, + -0.005856438539922237, + 0.0681174248456955, + 0.08307737112045288, + 0.0709967315196991, + 0.01765795797109604, + -0.006312967278063297, + 0.1759101301431656, + 0.11143764108419418, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.05968387424945831, + 0.04968845471739769, + 0.05649963393807411, + 0.0, + 0.0, + 0.0, + 0.007027155719697475, + 0.09547138959169388, + 0.0893152579665184, + 0.0841650441288948, + 0.010816042311489582, + 0.08700600266456604, + 0.08792556822299957, + 0.08387237787246704, + 0.007760591804981232, + 0.08358751982450485, + 0.08469780534505844, + 0.08339788019657135, + 0.023414811119437218, + 0.007603930775076151, + 0.1377788633108139, + 0.09705618768930435, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -0.018190324240759947, + 0.0, + 0.0, + 0.0, + -0.025669337847532007, + 0.0, + 0.0, + 0.0, + -0.02057921951431399, + 0.0, + 0.0, + 0.0, + -0.004167236053943634, + -0.02001873709484935, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + 0.0005509110837709184, + 0.36818173360824585, + 0.3681615619659424, + 0.36816001462936404, + -4.0280349494770107e-7, + 0.36818108654022214, + 0.36816150569915773, + 0.36816001462936404, + -9.199607329708759e-7, + 0.3681804394721985, + 0.3681614212989807, + 0.36816001462936404, + 0.08091667350530625, + 0.005898629778623581, + 0.3681445419073105, + 0.3680820146262646, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/RollingPin-v1/source_audit.json b/previous_single_task/RollingPin-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..2408a4aea24ee44a350fbc8d271647db5c160ae2 --- /dev/null +++ b/previous_single_task/RollingPin-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "RollingPin-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.9", + "peak_step": "35000", + "final": "0.75", + "start": "2026-08-21 13:44:37", + "eval_mtime": "2026-08-24 18:48:53.345574", + "source_mtime": "2026-05-14 07:01:36.021461", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/single_arm_dex/level_2/rooling_pin.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_RollingPin-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_RollingPin-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_RollingPin-v1/lora_params/lora_step_0035000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/RopePalmCageHoldDex-v1/eval_result.json b/previous_single_task/RopePalmCageHoldDex-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..631fcc368f9165c844c26052c6184da64509be5e --- /dev/null +++ b/previous_single_task/RopePalmCageHoldDex-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "RopePalmCageHoldDex-v1", + "level": "level_1", + "suite": "deformable/single_arm_dex", + "step": 5000, + "episodes": 50, + "successes": 47, + "success_rate": 0.94, + "return_mean": 58.918224368095395, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/RopePalmCageHoldDex-v1/evaluation_history.csv b/previous_single_task/RopePalmCageHoldDex-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..070b83219d8cd2ffd7b8db0f21fd095ac8c916ab --- /dev/null +++ b/previous_single_task/RopePalmCageHoldDex-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,47,0.94,58.918224368095395,True +10000,50,40,0.8,65.04999831333757,True +15000,50,44,0.88,60.85013444259763,True +20000,50,41,0.82,62.921965073496104,True +25000,50,42,0.84,63.20006484284997,True +30000,50,44,0.88,61.99557837814093,True +35000,50,44,0.88,62.34238040417433,True +40000,50,43,0.86,62.77147597208619,True diff --git a/previous_single_task/RopePalmCageHoldDex-v1/lora_step_0005000.npz b/previous_single_task/RopePalmCageHoldDex-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..bd392a810177b7e47b5978f5e6e84249616097f8 --- /dev/null +++ b/previous_single_task/RopePalmCageHoldDex-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:1ce86af8352f14a4738825f6407f5982bb1d7d8c6ece9030bbb98148f39fb05c +size 1867844866 diff --git a/previous_single_task/RopePalmCageHoldDex-v1/metadata.json b/previous_single_task/RopePalmCageHoldDex-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..a07c4af5464ea8bef49bbcc7e800b6e031f41378 --- /dev/null +++ b/previous_single_task/RopePalmCageHoldDex-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "RopePalmCageHoldDex-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "1ce86af8352f14a4738825f6407f5982bb1d7d8c6ece9030bbb98148f39fb05c", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "9cb9c54e126d6f121423d9f75adccaf2b1424af0112964ce2025f183cb1f066a", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_RopePalmCageHoldDex-v1", + "source_path": "maniladder/envs/deformable/single_arm_dex/level_1/rope_holding.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/RopePalmCageHoldDex-v1/norm_stats.json b/previous_single_task/RopePalmCageHoldDex-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..1021a71592ceb5b583a05d89e32e21da8801fbef --- /dev/null +++ b/previous_single_task/RopePalmCageHoldDex-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.006893170531839132, + -0.03138701617717743, + 0.009286044165492058, + 0.0, + 0.0, + 0.0, + -0.00026530143804848194, + 0.1091766282916069, + 0.08216720074415207, + 0.06112338975071907, + -0.00009108262020163238, + 0.041005294770002365, + 0.04081527143716812, + 0.04070103541016579, + -8.492523193126544e-6, + 0.04084920510649681, + 0.04076402261853218, + 0.040668241679668427, + 0.012749379500746727, + -0.006046037655323744, + 0.09204095602035522, + 0.07142707705497742, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.03461270034313202, + 0.046881262212991714, + 0.07405230402946472, + 0.0, + 0.0, + 0.0, + 0.0012236633338034153, + 0.14122018218040466, + 0.11162946373224258, + 0.09326664358377457, + 0.00014229454973246902, + 0.08083352446556091, + 0.08066096156835556, + 0.08058861643075943, + 0.0000357686985807959, + 0.08056993782520294, + 0.08059381693601608, + 0.08060141652822495, + 0.16998310387134552, + 0.00929404515773058, + 0.11803210526704788, + 0.09635671973228455, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + 0.0, + -0.0034370388112962245, + 0.0, + 0.0, + 0.0, + -0.0005754294518381357, + 0.0, + 0.0, + 0.0, + -0.00006913593342760578, + 0.0, + 0.0, + 0.0, + -0.1011815242767334, + -0.027238588999491185, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.038901058326661575, + 0.09996000148952008, + 0.0, + 0.0, + 0.0, + 0.002951310902833938, + 0.36528820509910587, + 0.359299666428566, + 0.3565380686581135, + -1.2531129177655205e-7, + 0.35597787292003635, + 0.3556280465900898, + 0.3555240412950516, + 0.0001859844467137009, + 0.3554005180478096, + 0.3554419126868248, + 0.3555240412950516, + 0.7655819606781006, + 0.000692458708118647, + 0.35547090108394624, + 0.3555244488120079, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/RopePalmCageHoldDex-v1/source_audit.json b/previous_single_task/RopePalmCageHoldDex-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..b6f362f5b8af34bad9a962d14494747052af8184 --- /dev/null +++ b/previous_single_task/RopePalmCageHoldDex-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "RopePalmCageHoldDex-v1", + "category": "reuse", + "level": "level_1", + "peak": "1.0", + "peak_step": "10000", + "final": "1.0", + "start": "2026-08-22 05:49:06", + "eval_mtime": "2026-08-25 04:09:54.336234", + "source_mtime": "2026-08-11 02:37:46.349079", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/single_arm_dex/level_1/rope_holding.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_RopePalmCageHoldDex-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_RopePalmCageHoldDex-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_RopePalmCageHoldDex-v1/lora_params/lora_step_0010000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/SpinPenHora-v1/eval_result.json b/previous_single_task/SpinPenHora-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..3e7d76e086c02edfed9fb7a69e1449f7112b4e96 --- /dev/null +++ b/previous_single_task/SpinPenHora-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "SpinPenHora-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_dex", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": 109.36473869925831, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/SpinPenHora-v1/evaluation_history.csv b/previous_single_task/SpinPenHora-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..466f02a095dd194c9015b77230a37ccd1f3ebd81 --- /dev/null +++ b/previous_single_task/SpinPenHora-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,109.36473869925831,True +10000,50,0,0.0,118.44748042443622,True +15000,50,0,0.0,104.3372119428232,True +20000,50,0,0.0,115.02319976602739,True +25000,50,0,0.0,85.35689656842325,True +30000,50,0,0.0,101.41080370466312,True +35000,50,0,0.0,110.3295705687703,True +40000,50,0,0.0,108.75212575052865,True diff --git a/previous_single_task/SpinPenHora-v1/lora_step_0005000.npz b/previous_single_task/SpinPenHora-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..5045a6af9d6fd083453360980acb1edcc00850ea --- /dev/null +++ b/previous_single_task/SpinPenHora-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:e325aba4c998d1b7e3725c8942e13a437e1aee71a5cb8738a2618f31a64e383b +size 1867844866 diff --git a/previous_single_task/SpinPenHora-v1/metadata.json b/previous_single_task/SpinPenHora-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..59afe741035a1e3977eedcfbbe6821230724507c --- /dev/null +++ b/previous_single_task/SpinPenHora-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "SpinPenHora-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "e325aba4c998d1b7e3725c8942e13a437e1aee71a5cb8738a2618f31a64e383b", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "9842fac20e6cfe33025d3de07ed1d493a510657545d719554beca688fdbcf2bc", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_SpinPenHora-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_4/spin_pen_hora.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/SpinPenHora-v1/norm_stats.json b/previous_single_task/SpinPenHora-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..fce21af622b6cb3ad5e897b60899ca9acca10665 --- /dev/null +++ b/previous_single_task/SpinPenHora-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.009335244074463844, + 0.0024976704735308886, + 0.007233715150505304, + -0.004378454759716988, + 0.002558615989983082, + -0.008510774932801723, + 0.004524897318333387, + 0.005287664942443371, + -0.004450587090104818, + -0.0044848560355603695, + -0.011426464654505253, + 0.002124013379216194, + -0.00561339920386672, + -0.005533966701477766, + -0.006719557102769613, + 0.006412609480321407, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.13271920382976532, + 0.0711321085691452, + 0.09883543103933334, + 0.05054338276386261, + 0.08974846452474594, + 0.08842353522777557, + 0.09289447218179703, + 0.08289776742458344, + 0.06917843222618103, + 0.1445661038160324, + 0.20448000729084015, + 0.0659499540925026, + 0.19341258704662323, + 0.18324744701385498, + 0.12188152968883514, + 0.11234397441148758, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.257874002635479, + -0.17787477800250057, + -0.19773405992984772, + -0.10273810553550722, + -0.18227622270584107, + -0.1721678829193115, + -0.20441216468811035, + -0.16306273794174195, + -0.1684794294834137, + -0.2089243411421776, + -0.3504902305603027, + -0.15985434055328368, + -0.4086200495958328, + -0.4242482316493988, + -0.26068603962063785, + -0.28645808038711545, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.2290968845844269, + 0.1546496360301971, + 0.20228927558660503, + 0.11905510163307187, + 0.17674527287483216, + 0.14802453994750975, + 0.17055097961425786, + 0.17329448747634885, + 0.1591359314918518, + 0.31670715073943134, + 0.3709222221374512, + 0.1443926706314087, + 0.4191422274708747, + 0.3428595760464669, + 0.22347704899907117, + 0.1766893914222718, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/SpinPenHora-v1/source_audit.json b/previous_single_task/SpinPenHora-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..7cdd8f33a75d5eb15176990160a6cbbc7d597098 --- /dev/null +++ b/previous_single_task/SpinPenHora-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "SpinPenHora-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-09-04 06:34:20", + "eval_mtime": "2026-09-07 08:48:24.389323", + "source_mtime": "2026-08-26 13:25:22.590266", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_4/spin_pen_hora.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_SpinPenHora-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_SpinPenHora-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_SpinPenHora-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/Stack6Blocks-v1/eval_result.json b/previous_single_task/Stack6Blocks-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..5ae6451af9ae74afedb31d4b2f24065a1799eabb --- /dev/null +++ b/previous_single_task/Stack6Blocks-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "Stack6Blocks-v1", + "level": "level_3", + "suite": "rigid_body/single_arm_gripper", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": 1412.5191994094848, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/Stack6Blocks-v1/evaluation_history.csv b/previous_single_task/Stack6Blocks-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..474fd09572d5f2fdd2b84ad04ca61ca710e334b2 --- /dev/null +++ b/previous_single_task/Stack6Blocks-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,1412.5191994094848,True +10000,50,0,0.0,1416.759504970312,True +15000,50,0,0.0,1410.7229782146214,True +20000,50,0,0.0,1404.6906222903729,True +25000,50,0,0.0,1405.1263322007655,True +30000,50,0,0.0,1418.389272699356,True +35000,50,0,0.0,1415.69502825737,True +40000,50,0,0.0,1405.1851704609394,True diff --git a/previous_single_task/Stack6Blocks-v1/lora_step_0005000.npz b/previous_single_task/Stack6Blocks-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..a0ff234c7b9d5db0b73e037e07a45d53620a4e2b --- /dev/null +++ b/previous_single_task/Stack6Blocks-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9ee601932f1d2357e3007185e9d991bf04d45e59123ba76950d9a70041c9c300 +size 1867844866 diff --git a/previous_single_task/Stack6Blocks-v1/metadata.json b/previous_single_task/Stack6Blocks-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..d640f5695665c904005ddd7b46cae14b886c61ca --- /dev/null +++ b/previous_single_task/Stack6Blocks-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "Stack6Blocks-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "9ee601932f1d2357e3007185e9d991bf04d45e59123ba76950d9a70041c9c300", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "1fb9df25a825d3954fa24577a691594532652ebbfbaabe6973d1fd0dd92ce651", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_Stack6Blocks-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_gripper/level_3/stack_blocks.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/Stack6Blocks-v1/norm_stats.json b/previous_single_task/Stack6Blocks-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..07774de2abfff24bf01e4cc3d54892843eee4d60 --- /dev/null +++ b/previous_single_task/Stack6Blocks-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.003235618118196726, + 0.0017587103648111224, + 0.00088964041788131, + 0.0, + 0.0, + 0.00003059045411646366, + -0.000520212750416249, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.032570965588092804, + 0.0532415509223938, + 0.0742519274353981, + 0.0, + 0.0, + 0.0031631223391741514, + 0.9999998211860657, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.10000000149011612, + -0.10000000149011612, + -0.10000000149011612, + 0.0, + 0.0, + -0.00006910889595747316, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.09996000148952008, + 0.09996000148952008, + 0.0, + 0.0, + -7.820768654340227e-6, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/Stack6Blocks-v1/source_audit.json b/previous_single_task/Stack6Blocks-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..812406f46f397dd4173fcb9d96e7655165d17bf6 --- /dev/null +++ b/previous_single_task/Stack6Blocks-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "Stack6Blocks-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 06:51:15", + "eval_mtime": "2026-08-25 13:13:47.423348", + "source_mtime": "2026-06-03 13:16:36.492484", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_gripper/level_3/stack_blocks.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_Stack6Blocks-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_Stack6Blocks-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_Stack6Blocks-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/StraightenTwoRopesSequential-v1/eval_result.json b/previous_single_task/StraightenTwoRopesSequential-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..01e7a133a8d92c9eee477aa0c15dd95ac6993b10 --- /dev/null +++ b/previous_single_task/StraightenTwoRopesSequential-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "StraightenTwoRopesSequential-v1", + "level": "level_2", + "suite": "deformable/single_arm_gripper", + "step": 15000, + "episodes": 50, + "successes": 18, + "success_rate": 0.36, + "return_mean": 1811.211635375023, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/StraightenTwoRopesSequential-v1/evaluation_history.csv b/previous_single_task/StraightenTwoRopesSequential-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..713ce4baedfa5b22d4913865e7c5b37dee2e6e65 --- /dev/null +++ b/previous_single_task/StraightenTwoRopesSequential-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,16,0.32,1831.5910737252236,True +10000,50,16,0.32,1816.7860989999772,True +15000,50,18,0.36,1811.211635375023,True +20000,50,10,0.2,2087.010603458881,True +25000,50,8,0.16,2118.8631678271295,True +30000,50,12,0.24,1981.979886984825,True +35000,50,14,0.28,1921.888679959774,True +40000,50,14,0.28,1892.6825481843948,True diff --git a/previous_single_task/StraightenTwoRopesSequential-v1/lora_step_0015000.npz b/previous_single_task/StraightenTwoRopesSequential-v1/lora_step_0015000.npz new file mode 100644 index 0000000000000000000000000000000000000000..57ff543d1d2e463cd1b962d14fe9ca8c32e232a3 --- /dev/null +++ b/previous_single_task/StraightenTwoRopesSequential-v1/lora_step_0015000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:9daa2193631b341750a5f0350bdc5e5649018b5fd2af5b2a8783086d4775d9d1 +size 1867844866 diff --git a/previous_single_task/StraightenTwoRopesSequential-v1/metadata.json b/previous_single_task/StraightenTwoRopesSequential-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..9753d3c882fa1159dbd588e928ed7d0eac631bee --- /dev/null +++ b/previous_single_task/StraightenTwoRopesSequential-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "StraightenTwoRopesSequential-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0015000.npz", + "checkpoint_sha256": "9daa2193631b341750a5f0350bdc5e5649018b5fd2af5b2a8783086d4775d9d1", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 15000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "f6e60a716ca1d4267dfd91c671f49c2d17fe5132404f4e6c41eca0e87e8949e6", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_StraightenTwoRopesSequential-v1", + "source_path": "maniladder/envs/deformable/single_arm_gripper/level_2/straighten_two_rope.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/StraightenTwoRopesSequential-v1/norm_stats.json b/previous_single_task/StraightenTwoRopesSequential-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..dfd29ca81a591707041e0f95f7174410f5035d4b --- /dev/null +++ b/previous_single_task/StraightenTwoRopesSequential-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.004165297374129295, + -0.008711563423275948, + -0.00635402649641037, + -0.002959220437332988, + -0.002409963170066476, + -0.003322085365653038, + 0.07704959809780121, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.09291274845600128, + 0.12684355676174164, + 0.10563207417726517, + 0.05974818393588066, + 0.062232378870248795, + 0.05997604876756668, + 0.9969637989997864, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.30000001192092896, + -0.30000001192092896, + -0.30000001192092896, + -0.15000000596046448, + -0.15000000596046448, + -0.15000000596046448, + -1.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.29988001191616065, + 0.29988001191616065, + 0.29988001191616065, + 0.14994000595808032, + 0.14994000595808032, + 0.14994000595808032, + 0.9996, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/StraightenTwoRopesSequential-v1/source_audit.json b/previous_single_task/StraightenTwoRopesSequential-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..5307ecb7921e7e8eced8d7c7b10dde266bb2122a --- /dev/null +++ b/previous_single_task/StraightenTwoRopesSequential-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "StraightenTwoRopesSequential-v1", + "category": "reuse", + "level": "level_2", + "peak": "0.3", + "peak_step": "10000", + "final": "0.05", + "start": "2026-09-04 06:14:34", + "eval_mtime": "2026-09-07 18:39:36.350654", + "source_mtime": "2026-08-23 04:17:50.013011", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/deformable/single_arm_gripper/level_2/straighten_two_rope.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_StraightenTwoRopesSequential-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_StraightenTwoRopesSequential-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_StraightenTwoRopesSequential-v1/lora_params/lora_step_0010000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/TennisRally-v1/eval_result.json b/previous_single_task/TennisRally-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..52d08b464dd6d3660f1a17499a4bfa0653e8c120 --- /dev/null +++ b/previous_single_task/TennisRally-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "TennisRally-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_dex", + "step": 15000, + "episodes": 50, + "successes": 3, + "success_rate": 0.06, + "return_mean": 631.176888383422, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/TennisRally-v1/evaluation_history.csv b/previous_single_task/TennisRally-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..70950038c4bbbf12c900d88a82d0e8965f4ab30f --- /dev/null +++ b/previous_single_task/TennisRally-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,1,0.02,357.17158117782327,True +10000,50,1,0.02,582.0495989676378,True +15000,50,3,0.06,631.176888383422,True +20000,50,1,0.02,664.8420089965593,True +25000,50,3,0.06,660.3962393769901,True +30000,50,2,0.04,564.6722093473934,True +35000,50,0,0.0,592.2262566924281,True +40000,50,3,0.06,668.7410189095605,True diff --git a/previous_single_task/TennisRally-v1/lora_step_0015000.npz b/previous_single_task/TennisRally-v1/lora_step_0015000.npz new file mode 100644 index 0000000000000000000000000000000000000000..a941a3d51706807d56c249d9db3e3c21cb16aded --- /dev/null +++ b/previous_single_task/TennisRally-v1/lora_step_0015000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:2092beb4bec50f1e469c47e456d7589d4b2dc0ae76890845ed06af1de8879b87 +size 1867844866 diff --git a/previous_single_task/TennisRally-v1/metadata.json b/previous_single_task/TennisRally-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..5af265702bd4d2c88404c7efa667cc9d7ea95e23 --- /dev/null +++ b/previous_single_task/TennisRally-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "TennisRally-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0015000.npz", + "checkpoint_sha256": "2092beb4bec50f1e469c47e456d7589d4b2dc0ae76890845ed06af1de8879b87", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 15000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "ca766944a0ac6dec27bc476a7d80846a569360f7e8441093332e5fc86df0aa2a", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_TennisRally-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_4/tennis_rally.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/TennisRally-v1/norm_stats.json b/previous_single_task/TennisRally-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..077acfc5688b35339d5548c2471324933f5d6a9b --- /dev/null +++ b/previous_single_task/TennisRally-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.013312429189682007, + 0.020284662023186684, + 0.014531883411109447, + 0.004442907404154539, + 0.000726330210454762, + -0.0022996775805950165, + -0.0690263882279396, + 0.325010746717453, + 0.213121235370636, + 0.09750287234783173, + -0.03455479443073273, + 0.2795143127441406, + 0.20858986675739288, + 0.10795512050390244, + 0.007813812233507633, + 0.19834384322166443, + 0.23197120428085327, + 0.15333771705627441, + 0.07417537271976471, + 0.10920114070177078, + 0.5167955756187439, + 0.14892138540744781, + 0.026550747454166412, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.050910819321870804, + 0.08863326907157898, + 0.11325422674417496, + 0.02185591496527195, + 0.006034791469573975, + 0.015843210741877556, + 0.043486788868904114, + 0.16701072454452515, + 0.12472402304410934, + 0.10498826950788498, + 0.02716856077313423, + 0.1501808911561966, + 0.1249406635761261, + 0.10472111403942108, + 0.008431053720414639, + 0.1321910321712494, + 0.13060154020786285, + 0.10582374036312103, + 0.0543028749525547, + 0.07820308953523636, + 0.2514328360557556, + 0.11867304146289825, + 0.16076630353927612, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.14355009770244365, + -0.050144569242000586, + -0.050084953510761254, + -6.120830308645707e-6, + -0.000023538317345082535, + -0.11999999731779099, + -0.23339781792306166, + -0.000017061927792383358, + -2.135506065314985e-6, + -9.932551447988658e-10, + -0.06787121084518731, + -0.000017738240785547532, + -2.2201229512575082e-6, + -9.930597455465318e-10, + -0.017667821326479316, + -0.0000184138843906112, + -2.3040986434352817e-6, + -9.928022848271212e-10, + 0.001023896577837877, + -0.028889524440467357, + 0.00007245183223858476, + 0.00007106602424755692, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.04985235320776704, + 0.33682937905788424, + 0.89976469014287, + 0.11997555615319871, + 0.012809324079658835, + 0.0019343491662293655, + -0.000052532581282127566, + 0.7012652069351752, + 0.653182160561498, + 0.6065280238803574, + 0.009970135086216025, + 0.6998244493328762, + 0.6531820953941307, + 0.6065280238804008, + 0.01608392636068165, + 0.6998236965383055, + 0.6531819840207933, + 0.6065280238804581, + 0.28510479467404076, + 0.23368673821166158, + 0.7401448540470912, + 0.5856182017108542, + 0.9998, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/TennisRally-v1/source_audit.json b/previous_single_task/TennisRally-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..1660687b3890145bd584ce16c840196d9d6d1110 --- /dev/null +++ b/previous_single_task/TennisRally-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "TennisRally-v1", + "category": "reuse", + "level": "level_4", + "peak": "0.1", + "peak_step": "15000", + "final": "0.0", + "start": "2026-08-22 06:11:25", + "eval_mtime": "2026-08-25 06:01:09.263282", + "source_mtime": "2026-08-09 10:00:46.699489", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_4/tennis_rally.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_TennisRally-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_TennisRally-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_TennisRally-v1/lora_params/lora_step_0015000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/TurnOnFaucet-v1/eval_result.json b/previous_single_task/TurnOnFaucet-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..8f5650a6e9567d225ae8cae73d01667aeaf7c2cc --- /dev/null +++ b/previous_single_task/TurnOnFaucet-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "TurnOnFaucet-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_dex", + "step": 10000, + "episodes": 50, + "successes": 46, + "success_rate": 0.92, + "return_mean": 16.862867071165237, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/TurnOnFaucet-v1/evaluation_history.csv b/previous_single_task/TurnOnFaucet-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..620a98705d477d909ab645162e9477e04244ab6c --- /dev/null +++ b/previous_single_task/TurnOnFaucet-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,42,0.84,14.629086217982696,True +10000,50,46,0.92,16.862867071165237,True +15000,50,43,0.86,15.642866224958563,True +20000,50,44,0.88,15.11060263212072,True +25000,50,43,0.86,15.139536786193503,True +30000,50,41,0.82,14.144072188258878,True +35000,50,41,0.82,14.374822221453078,True +40000,50,42,0.84,14.696782298122711,True diff --git a/previous_single_task/TurnOnFaucet-v1/lora_step_0010000.npz b/previous_single_task/TurnOnFaucet-v1/lora_step_0010000.npz new file mode 100644 index 0000000000000000000000000000000000000000..d6eacb29d720cc74a2438e92aa7758c822c9d671 --- /dev/null +++ b/previous_single_task/TurnOnFaucet-v1/lora_step_0010000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:c233e2a4e9a384ecc107c403938aebb21fc757cc675015d293b96ffd0fce91c0 +size 1867844866 diff --git a/previous_single_task/TurnOnFaucet-v1/metadata.json b/previous_single_task/TurnOnFaucet-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..dc0a2f8e5656318b4d2fa719476da67af46e147c --- /dev/null +++ b/previous_single_task/TurnOnFaucet-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "TurnOnFaucet-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0010000.npz", + "checkpoint_sha256": "c233e2a4e9a384ecc107c403938aebb21fc757cc675015d293b96ffd0fce91c0", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 10000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "d7ff021a8c239146a9674ae42531575ec2779038adf1b62cece7d25476fe1092", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_TurnOnFaucet-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_1/turn_on_faucet.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/TurnOnFaucet-v1/norm_stats.json b/previous_single_task/TurnOnFaucet-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..074e0c5e34fbaf9952b08fd32bcbaa3f5dfc249e --- /dev/null +++ b/previous_single_task/TurnOnFaucet-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + 0.004291241522878408, + -0.042976848781108856, + 0.0028307437896728516, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.02202862873673439, + 0.04887108877301216, + 0.013742905110120773, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.027680000412464142, + -0.10000000149011612, + -0.000012363001704215995, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.09996000148952008, + 0.10000000149011612, + 0.10000000149011612, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/TurnOnFaucet-v1/source_audit.json b/previous_single_task/TurnOnFaucet-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..4943e368b1c04259486fc04ecc50228ed983e78f --- /dev/null +++ b/previous_single_task/TurnOnFaucet-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "TurnOnFaucet-v1", + "category": "reuse", + "level": "level_1", + "peak": "0.95", + "peak_step": "40000", + "final": "0.95", + "start": "2026-08-21 13:37:15", + "eval_mtime": "2026-08-24 10:54:34.074237", + "source_mtime": "2026-02-18 07:42:17.449436", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_1/turn_on_faucet.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_TurnOnFaucet-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_TurnOnFaucet-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_TurnOnFaucet-v1/lora_params/lora_step_0040000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/TweezersPickPlace-v1/eval_result.json b/previous_single_task/TweezersPickPlace-v1/eval_result.json new file mode 100644 index 0000000000000000000000000000000000000000..705594ab906ed533f7180df43e3549ec39da8e52 --- /dev/null +++ b/previous_single_task/TweezersPickPlace-v1/eval_result.json @@ -0,0 +1,17 @@ +{ + "task": "TweezersPickPlace-v1", + "level": "level_3", + "suite": "rigid_body/single_arm_dex", + "step": 5000, + "episodes": 50, + "successes": 0, + "success_rate": 0.0, + "return_mean": -150.75219471901656, + "return_mean_finite": true, + "episode_seed_base": 1000, + "ckpts_evaluated": 8, + "selection": "highest available evaluated success rate; earliest step breaks ties", + "evaluation_type": "offline_eval50", + "individual_episode_traces_included": false, + "note": "Original single-task training run; not retrained in the current campaign." +} diff --git a/previous_single_task/TweezersPickPlace-v1/evaluation_history.csv b/previous_single_task/TweezersPickPlace-v1/evaluation_history.csv new file mode 100644 index 0000000000000000000000000000000000000000..459e77ac7116ed1dc5ad92507f6e63d626cde0dd --- /dev/null +++ b/previous_single_task/TweezersPickPlace-v1/evaluation_history.csv @@ -0,0 +1,9 @@ +step,episodes,successes,success_rate,return_mean,return_mean_finite +5000,50,0,0.0,-150.75219471901656,True +10000,50,0,0.0,-149.61620528012514,True +15000,50,0,0.0,-144.82497310459613,True +20000,50,0,0.0,-149.89631738722323,True +25000,50,0,0.0,-148.6842001259327,True +30000,50,0,0.0,-135.40562374919654,True +35000,50,0,0.0,-147.4044314825535,True +40000,50,0,0.0,-141.0583949536085,True diff --git a/previous_single_task/TweezersPickPlace-v1/lora_step_0005000.npz b/previous_single_task/TweezersPickPlace-v1/lora_step_0005000.npz new file mode 100644 index 0000000000000000000000000000000000000000..b3b72357fd5a648f4682e186c4110272365d9152 --- /dev/null +++ b/previous_single_task/TweezersPickPlace-v1/lora_step_0005000.npz @@ -0,0 +1,3 @@ +version https://git-lfs.github.com/spec/v1 +oid sha256:23135a349b12ab66e78abcd43364a7f1c0f6a05b13ae08bc8cfb626005f6d306 +size 1867844866 diff --git a/previous_single_task/TweezersPickPlace-v1/metadata.json b/previous_single_task/TweezersPickPlace-v1/metadata.json new file mode 100644 index 0000000000000000000000000000000000000000..82b31c61861432365e81af1537a7dd7dfcd11b9e --- /dev/null +++ b/previous_single_task/TweezersPickPlace-v1/metadata.json @@ -0,0 +1,30 @@ +{ + "task": "TweezersPickPlace-v1", + "method": "PI0.5 LoRA + trainable weights", + "base_checkpoint": "gs://openpi-assets/checkpoints/pi05_base/params", + "checkpoint": "lora_step_0005000.npz", + "checkpoint_sha256": "23135a349b12ab66e78abcd43364a7f1c0f6a05b13ae08bc8cfb626005f6d306", + "checkpoint_bytes": 1867844866, + "tensor_count": 51, + "action_dim": 32, + "state_dim": 32, + "action_horizon": 50, + "max_token_len": 200, + "image_size": 224, + "exec_steps": 16, + "sample_steps": 10, + "state_input": "zero vector", + "control_mode": "pd_ee_delta_pose", + "num_demos": 20, + "train_steps": 40000, + "batch_size": 64, + "selected_step": 5000, + "eval_episodes": 50, + "eval_seed_base": 1000, + "norm_stats_sha256": "0f0f5f9ad2ab9f1b79251bc8ca83acbf346fc12435b3c57c8834059697969e73", + "norm_stats_file": "norm_stats.json", + "training_run": "pi05_lora_task_TweezersPickPlace-v1", + "source_path": "maniladder/envs/rigid_body/single_arm_dex/level_3/tweezers_pick_place.py", + "environment_audit": "Previously classified reusable by the 2026-10-22 audit.", + "environment_audit_limit": "Training did not preserve a complete source snapshot. Reuse was inferred from source timestamps and change review; exact source identity is not proven." +} diff --git a/previous_single_task/TweezersPickPlace-v1/norm_stats.json b/previous_single_task/TweezersPickPlace-v1/norm_stats.json new file mode 100644 index 0000000000000000000000000000000000000000..e16b1f43a0809e0a494a690f224928d602674ed7 --- /dev/null +++ b/previous_single_task/TweezersPickPlace-v1/norm_stats.json @@ -0,0 +1,280 @@ +{ + "norm_stats": { + "state": { + "mean": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + }, + "actions": { + "mean": [ + -0.005864385981112719, + 0.013714548200368881, + -0.007069722283631563, + -0.0023021153174340725, + 0.0, + 0.0, + -0.010505079291760921, + 0.10787301510572433, + 0.05691135674715042, + 0.033528201282024384, + -0.0055929189547896385, + 0.053711898624897, + 0.05497199669480324, + 0.027210645377635956, + -0.01281710248440504, + 0.051465459167957306, + 0.03430143743753433, + 0.013513023965060711, + 0.012894163839519024, + 0.007050583139061928, + 0.10843048989772797, + 0.03423301875591278, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "std": [ + 0.0457100011408329, + 0.10249485075473785, + 0.09771070629358292, + 0.02387947030365467, + 0.0, + 0.0, + 0.01627831906080246, + 0.10885902494192123, + 0.07467794418334961, + 0.06134865805506706, + 0.004575683735311031, + 0.0714765340089798, + 0.08059069514274597, + 0.06506350636482239, + 0.012757747434079647, + 0.08178473263978958, + 0.07021723687648773, + 0.06687481701374054, + 0.08723316341638565, + 0.01909809745848179, + 0.1355769783258438, + 0.0741838812828064, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q01": [ + -0.15000000596046448, + -0.15000000596046448, + -0.15000000596046448, + -0.00004999999999999449, + 0.0, + 0.0, + -0.05607390509545803, + -0.08844716250896453, + -0.11948952668905258, + -0.13133321504592896, + -0.013014046675626742, + -0.11871483231186866, + -0.11502016551494598, + -0.1358983793258667, + -0.03389143269928172, + -0.12270068773031234, + -0.13359320558905602, + -0.14784693579673766, + -0.12825568587779998, + -0.03561986461281776, + -0.10256628575325012, + -0.14303029563426972, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ], + "q99": [ + 0.14994000595808032, + 0.14994000595808032, + 0.14994000595808032, + -0.00004999999999999449, + 0.0, + 0.0, + 0.03732366542518138, + 0.34413003623485566, + 0.2925756798863411, + 0.20101750946044922, + 6.838083332330949e-6, + 0.2502417660534382, + 0.3134013623952866, + 0.2299366989612579, + 0.0030938232460059195, + 0.3728567148923875, + 0.30175620855093, + 0.23867867076992988, + 0.41050759950876237, + 0.07123147225379944, + 0.5531813698351383, + 0.2775846379995346, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0, + 0.0 + ] + } + } +} \ No newline at end of file diff --git a/previous_single_task/TweezersPickPlace-v1/source_audit.json b/previous_single_task/TweezersPickPlace-v1/source_audit.json new file mode 100644 index 0000000000000000000000000000000000000000..520da6f3cbc77cc07556347994948a51ac184771 --- /dev/null +++ b/previous_single_task/TweezersPickPlace-v1/source_audit.json @@ -0,0 +1,19 @@ +{ + "task": "TweezersPickPlace-v1", + "category": "reuse", + "level": "level_3", + "peak": "0.0", + "peak_step": "5000", + "final": "0.0", + "start": "2026-08-22 06:11:28", + "eval_mtime": "2026-08-25 07:41:26.992702", + "source_mtime": "2026-05-25 04:48:52.295277", + "reason": "\u73b0\u5b58task\u6e90\u7801\u6700\u540e\u4fee\u6539\u65e9\u4e8e\u672c\u6b21\u8bad\u7ec3\uff0c\u672a\u53d1\u73b0\u540e\u7eed\u4efb\u52a1\u8bed\u4e49\u53d8\u5316\uff1b9\u67089\u65e5\u96c6\u4e2d\u63d0\u4ea4\u4e0d\u5355\u72ec\u89c6\u4e3a\u6539\u52a8\u65f6\u95f4", + "path": "maniladder/envs/rigid_body/single_arm_dex/level_3/tweezers_pick_place.py", + "eval_log": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_TweezersPickPlace-v1/eval_log.txt", + "train_log": "/mnt/gck/maniladder_logs/pi05_tasks/train_TweezersPickPlace-v1.log", + "peak_snapshot": "/mnt/gck/openpi_checkpoints_tasks/maniladder_pi05_lora/pi05_lora_task_TweezersPickPlace-v1/lora_params/lora_step_0005000.npz", + "peak_snapshot_exists": "True", + "complete_40000": "True", + "evidence_limit": "\u6e90\u7801mtime\u8f85\u52a9\u63a8\u65ad\uff1brun\u672a\u4fdd\u5b58\u5b8c\u6574\u6e90\u7801/git diff\uff0c\u4e0d\u80fd\u4ec5\u7531\u96c6\u4e2d\u63d0\u4ea4\u65f6\u95f4\u8fd8\u539f\u8bad\u7ec3\u6e90\u7801" +} diff --git a/previous_single_task/manifest.csv b/previous_single_task/manifest.csv new file mode 100644 index 0000000000000000000000000000000000000000..0fc517217c482ed3d819c7e207457a8143ba202b --- /dev/null +++ b/previous_single_task/manifest.csv @@ -0,0 +1,63 @@ +task,level,suite,step,success_rate,successes,episodes,ckpts_evaluated,checkpoint,checkpoint_sha256,checkpoint_bytes +DualArmCoverCloth-v1,level_1,deformable/dual_arm_dex,15000,0.8,40,50,8,previous_single_task/DualArmCoverCloth-v1/lora_step_0015000.npz,05fe236e482b4a9e6046ac1448bfe952538ed4b7eb217a0961a0e5228f4c91f4,1867943218 +DualArmRopeStretch-v1,level_1,deformable/dual_arm_dex,10000,0.92,46,50,8,previous_single_task/DualArmRopeStretch-v1/lora_step_0010000.npz,50616181d63062ba47a5a893add02f9b6a99743801d4690f62f11ecb1250feda,1867943218 +DualArmFoldHandKerchief-v1,level_1,deformable/dual_arm_gripper,5000,0.74,37,50,8,previous_single_task/DualArmFoldHandKerchief-v1/lora_step_0005000.npz,aae641e518457c83346e4ba2c617c3175eb9a18b6d8e3f2c526b20de921868aa,1867844866 +DualArmPouringWater-v1,level_1,deformable/dual_arm_gripper,30000,0.95,19,20,8,previous_single_task/DualArmPouringWater-v1/lora_step_0030000.npz,d1a0affb3e58bc3eb0b23f5ae83cb9a7f64a8be66940868409a87cb0b1b4ca34,1867844866 +DualArmRopeCylinderWarpping-v1,level_1,deformable/dual_arm_gripper,10000,0.62,31,50,8,previous_single_task/DualArmRopeCylinderWarpping-v1/lora_step_0010000.npz,a75996da6ae851c12335001a14a54ce38cf3038772fc242de85651ecd4992c6a,1867844866 +RopePalmCageHoldDex-v1,level_1,deformable/single_arm_dex,5000,0.94,47,50,8,previous_single_task/RopePalmCageHoldDex-v1/lora_step_0005000.npz,1ce86af8352f14a4738825f6407f5982bb1d7d8c6ece9030bbb98148f39fb05c,1867844866 +DualArmOpenPen-v1,level_1,rigid_body/dual_arm_dex,35000,0.36,18,50,8,previous_single_task/DualArmOpenPen-v1/lora_step_0035000.npz,8057718dae051ffe7dd5641fc5e50f135068359806d80cd29a273349c008dd78,1867943218 +DualArmButtonPress-v1,level_1,rigid_body/dual_arm_gripper,5000,0.98,49,50,8,previous_single_task/DualArmButtonPress-v1/lora_step_0005000.npz,fe6b5128698f903f0b7a412f09be569043fdb7f2ae8cb8eb2edc016ff77b1111,1867844866 +GraspCylinder-v1,level_1,rigid_body/single_arm_dex,15000,0.94,47,50,8,previous_single_task/GraspCylinder-v1/lora_step_0015000.npz,8df817057bb6ac221d574592a7437a1607aae1853e0350e3426507215748d225,1867844866 +LiftBucket-v1,level_1,rigid_body/single_arm_dex,40000,0.96,48,50,8,previous_single_task/LiftBucket-v1/lora_step_0040000.npz,5040809ab7b421527b71dce655b290016467761ca5460a43a32fa2121d2519ee,1867844866 +TurnOnFaucet-v1,level_1,rigid_body/single_arm_dex,10000,0.92,46,50,8,previous_single_task/TurnOnFaucet-v1/lora_step_0010000.npz,c233e2a4e9a384ecc107c403938aebb21fc757cc675015d293b96ffd0fce91c0,1867844866 +OpenDoor-v1,level_1,rigid_body/single_arm_gripper,5000,0.3,15,50,8,previous_single_task/OpenDoor-v1/lora_step_0005000.npz,9becc4d5effe10eca7baef2875443f2e5ab55c21ec18fbc863537e30280973f7,1867844866 +PickPlacePen-v1,level_1,rigid_body/single_arm_gripper,20000,0.36,18,50,8,previous_single_task/PickPlacePen-v1/lora_step_0020000.npz,8969ee8461c5a3ce36cb0a205474c1214990589054aa85ce2f3b1f707e1ab16f,1867844866 +PushSoccer-v1,level_1,rigid_body/single_arm_gripper,40000,0.8,40,50,8,previous_single_task/PushSoccer-v1/lora_step_0040000.npz,b495183247ad15943fb3abd608c423928d8ceba2859efef3380fcc4818297a54,1867844866 +DualArmPlaceRopeOverRod-v1,level_2,deformable/dual_arm_dex,40000,0.46,23,50,8,previous_single_task/DualArmPlaceRopeOverRod-v1/lora_step_0040000.npz,ab110fb81eabcc3bd39bb226cd5856d3b8be169e6ec2745d0b8aebaa94d2ac5e,1867943218 +DualArmStuffCloth-v1,level_2,deformable/dual_arm_dex,15000,0.4,20,50,8,previous_single_task/DualArmStuffCloth-v1/lora_step_0015000.npz,727b90f7ce8ebdf643ce17e321bf0546117642cc589645f96c2c941e6bb1a7a9,1867943218 +DualArmClothAlign-v1,level_2,deformable/dual_arm_gripper,40000,0.38,19,50,8,previous_single_task/DualArmClothAlign-v1/lora_step_0040000.npz,c17241b647ad68f11aacc420f950dc5333abd47e7e665e9c0dc84af567943bac,1867844866 +ClothWipe-v1,level_2,deformable/single_arm_dex,25000,0.52,26,50,8,previous_single_task/ClothWipe-v1/lora_step_0025000.npz,1603d8b01f721d5ea488ada553dc31fcfad4ac54785d00a639f3d9243e13ae26,1867844866 +RollingPin-v1,level_2,deformable/single_arm_dex,25000,0.96,48,50,8,previous_single_task/RollingPin-v1/lora_step_0025000.npz,49c64d6d95014fb02dc5fde1dd3adfac52c28264a1fd6a7a7cd8499a06f410ac,1867844866 +FoldShirt-v1,level_2,deformable/single_arm_gripper,40000,0.72,36,50,8,previous_single_task/FoldShirt-v1/lora_step_0040000.npz,594d272a50755c96e1a9cd31a8a961962a4b34c7295a64fc39ed716a4e19cbe0,1867844866 +StraightenTwoRopesSequential-v1,level_2,deformable/single_arm_gripper,15000,0.36,18,50,8,previous_single_task/StraightenTwoRopesSequential-v1/lora_step_0015000.npz,9daa2193631b341750a5f0350bdc5e5649018b5fd2af5b2a8783086d4775d9d1,1867844866 +DualArmOpenBottle-v1,level_2,rigid_body/dual_arm_dex,30000,0.06,3,50,8,previous_single_task/DualArmOpenBottle-v1/lora_step_0030000.npz,badce300065df7e4c45bb4cfb07dfcf74dcd6a33c524f6de76ef3a15aff2490f,1867943218 +DualArmOpenLaptop-v1,level_2,rigid_body/dual_arm_dex,25000,0.42,21,50,8,previous_single_task/DualArmOpenLaptop-v1/lora_step_0025000.npz,96e016a4d0628be868f5ae1108397addc4216fb3123020171775346ad335cca3,1867943218 +DualArmThrowBlock-v1,level_2,rigid_body/dual_arm_dex,5000,0.0,0,50,8,previous_single_task/DualArmThrowBlock-v1/lora_step_0005000.npz,ddc085746228651a8ef69d51c548e10e6c50ca76081808df23890913a3bf20e0,1867943218 +DualArmPegInHole-v1,level_2,rigid_body/dual_arm_gripper,5000,0.18,9,50,8,previous_single_task/DualArmPegInHole-v1/lora_step_0005000.npz,3da1d0202466bf4a0182a47ddb00ecd42112a22ed462c85197a9b5625f4d5508,1867844866 +GraspTwoBlocks-v1,level_2,rigid_body/single_arm_dex,5000,0.72,36,50,8,previous_single_task/GraspTwoBlocks-v1/lora_step_0005000.npz,84de21676c4ded27f090d19e152584578a07ce0de6816d732e84f967a9378831,1867844866 +HammerNailOut-v1,level_2,rigid_body/single_arm_dex,40000,0.44,22,50,8,previous_single_task/HammerNailOut-v1/lora_step_0040000.npz,8dc1babeb6ad28356e7a4fedc58122db41c17d4564e0ac1d95cd7c8fbe5b228a,1867844866 +ReorientPen-v1,level_2,rigid_body/single_arm_dex,25000,0.38,19,50,8,previous_single_task/ReorientPen-v1/lora_step_0025000.npz,070504c2d43469bc87fd254b20ab68c2c4b646c9acf7fc066a6a5ad0e5bde174,1867844866 +HangMug-v1,level_2,rigid_body/single_arm_gripper,10000,0.7,35,50,8,previous_single_task/HangMug-v1/lora_step_0010000.npz,7ad647a1261390f68a70fa7320c3238a05aec39db9d37d8a34be76c63ee1c8ed,1867844866 +KitAssembly-v1,level_2,rigid_body/single_arm_gripper,15000,0.06,3,50,8,previous_single_task/KitAssembly-v1/lora_step_0015000.npz,2e78e125fcb2aacf7a1c59973868de4839dc8bf7c6fdfa649b5ebde9b3841189,1867844866 +PivotBook-v1,level_2,rigid_body/single_arm_gripper,15000,0.38,19,50,8,previous_single_task/PivotBook-v1/lora_step_0015000.npz,1f419832345ac22ea2f4a05686c32c7670f62d2704c5b5025e19414cfa51b047,1867844866 +DualArmShapeDoll-v1,level_3,deformable/dual_arm_gripper,30000,0.04,2,50,8,previous_single_task/DualArmShapeDoll-v1/lora_step_0030000.npz,e290f251fb1feb05077deb5167416a0bf34c273229cff2ad60b0dfbd91934955,1867844866 +DualArmThreadRopeThroughTube-v1,level_3,deformable/dual_arm_gripper,35000,0.16,8,50,8,previous_single_task/DualArmThreadRopeThroughTube-v1/lora_step_0035000.npz,ed640c43be58457fab4969a591ef4ab14103d5553c3f363f39412e0ff47bbdd1,1867844866 +DualArmWeaveCloth-v1,level_3,deformable/dual_arm_gripper,20000,0.32,16,50,8,previous_single_task/DualArmWeaveCloth-v1/lora_step_0020000.npz,0db96e943ccf5a38c41f132f289033e36e87358deb637445b03c9fdc37a1bd60,1867844866 +HitBox-v1,level_3,deformable/single_arm_dex,25000,0.06,3,50,8,previous_single_task/HitBox-v1/lora_step_0025000.npz,fc4b6cdf6398e6b7222b8a9ba80630fbd262fb2de923829c55aa3aa1c2df27c0,1867844866 +LiquidMeasuring-v1,level_3,deformable/single_arm_gripper,35000,0.32,16,50,8,previous_single_task/LiquidMeasuring-v1/lora_step_0035000.npz,6e0a9a12b8f5882b391f082b0640baae09dcab280030c3e437804a56f1fa7a8f,1867844866 +DualArmPingpongBalance-v1,level_3,rigid_body/dual_arm_dex,15000,0.2,10,50,8,previous_single_task/DualArmPingpongBalance-v1/lora_step_0015000.npz,da1fb459a8bcc79cb0985cb6e78eb4e477f7c2a88908422063379612734dab5d,1867943218 +DualArmSetupTable-v1,level_3,rigid_body/dual_arm_dex,25000,0.04,2,50,8,previous_single_task/DualArmSetupTable-v1/lora_step_0025000.npz,093e094f8ea395f959f4160b04c57930e33c5a3df4cc6504428a877b03a2244c,1867992394 +DualArmOpenFridge-v1,level_3,rigid_body/dual_arm_gripper,5000,0.0,0,50,8,previous_single_task/DualArmOpenFridge-v1/lora_step_0005000.npz,311033395a4c57babcae115e5a0ae2c179504482d1d85075956fd03d545c5e48,1867844866 +HammerNail-v1,level_3,rigid_body/single_arm_dex,35000,0.12,6,50,8,previous_single_task/HammerNail-v1/lora_step_0035000.npz,a26699d0543203305445941f7008a0c0ed01afa2b451ffa3850d54fbc45f4dc9,1867844866 +MobilePickPlace-v1,level_3,rigid_body/single_arm_dex,15000,0.08,4,50,8,previous_single_task/MobilePickPlace-v1/lora_step_0015000.npz,a5f73d09c256f142d64b4edf03062bc328f6a3d0c13c85a0949a04d875cd64cb,1867844866 +TweezersPickPlace-v1,level_3,rigid_body/single_arm_dex,5000,0.0,0,50,8,previous_single_task/TweezersPickPlace-v1/lora_step_0005000.npz,23135a349b12ab66e78abcd43364a7f1c0f6a05b13ae08bc8cfb626005f6d306,1867844866 +MazeBlockPush-v1,level_3,rigid_body/single_arm_gripper,30000,0.65,13,20,8,previous_single_task/MazeBlockPush-v1/lora_step_0030000.npz,e7396df9ba9f96b5083f9ca0b1e128b7ce0bc88b8eb5319987eb033f60269602,1867844866 +Stack6Blocks-v1,level_3,rigid_body/single_arm_gripper,5000,0.0,0,50,8,previous_single_task/Stack6Blocks-v1/lora_step_0005000.npz,9ee601932f1d2357e3007185e9d991bf04d45e59123ba76950d9a70041c9c300,1867844866 +DualArmJumpRope-v1,level_4,deformable/dual_arm_dex,20000,0.02,1,50,8,previous_single_task/DualArmJumpRope-v1/lora_step_0020000.npz,15b86fb5ec34e19d2161c8f535d0ba3c2d0917f4b1434d574d1bb56ee7367487,1867943218 +DualArmClothHanger-v1,level_4,deformable/dual_arm_gripper,5000,0.0,0,50,8,previous_single_task/DualArmClothHanger-v1/lora_step_0005000.npz,7c5dae817921bdcdf2474ea67dc5e7edd59fb41832db3e06af5ffc9f1f60c28c,1867844866 +BuildTent-v1,level_4,deformable/single_arm_gripper,10000,0.06,3,50,8,previous_single_task/BuildTent-v1/lora_step_0010000.npz,d27916914df184fddd0df81bde6d7f7187f76d203648989b74d2711fca67281d,1867844866 +DualArmFlipTable-v1,level_4,rigid_body/dual_arm_dex,5000,0.0,0,50,8,previous_single_task/DualArmFlipTable-v1/lora_step_0005000.npz,84f91077bad9c28bf3fd78e16d44e41b3de8a988f36eea6b6aa3346e44f7790a,1867943218 +DualArmKeyboardType-v1,level_4,rigid_body/dual_arm_dex,5000,0.0,0,50,7,previous_single_task/DualArmKeyboardType-v1/lora_step_0005000.npz,521d586261c29c45cf239aebd59190737d7be7d05cc1afd943f361e72fa76bee,1867943218 +DualArmSendPingpong-v1,level_4,rigid_body/dual_arm_dex,40000,0.08,4,50,8,previous_single_task/DualArmSendPingpong-v1/lora_step_0040000.npz,70924c6c7880fc34d61197d289cbbbbdc15a3e146339aed6b10af36f19ea3126,1867943218 +DualArmShootBasketball-v1,level_4,rigid_body/dual_arm_dex,5000,0.0,0,50,8,previous_single_task/DualArmShootBasketball-v1/lora_step_0005000.npz,03ce4d67db10529e152c37e5894df057c430f38592273d6ec47cbbd4500226e7,1867992394 +DualArmBallClassify-v1,level_4,rigid_body/dual_arm_gripper,10000,0.0,0,50,7,previous_single_task/DualArmBallClassify-v1/lora_step_0010000.npz,0bec5644e89c77f1fbdb520eae5ea4504903d0bbed7dab3b41295d045e50aa0f,1867844866 +DualArmBilliardBridge-v1,level_4,rigid_body/dual_arm_gripper,5000,0.0,0,50,8,previous_single_task/DualArmBilliardBridge-v1/lora_step_0005000.npz,7dabff95c821665a5f6ad6d381ca64ce8dc8ae1baa11aa6b1cc042d3ebc6bc9e,1867844866 +DualArmMahjongSetup-v1,level_4,rigid_body/dual_arm_gripper,5000,0.02,1,50,8,previous_single_task/DualArmMahjongSetup-v1/lora_step_0005000.npz,f565771d810c4e9c7d1499feee6ffbe540235cabab920b80634365b74c11a51a,1867844866 +FitnessBall-v1,level_4,rigid_body/single_arm_dex,5000,0.0,0,50,8,previous_single_task/FitnessBall-v1/lora_step_0005000.npz,063779d861fa3fb188065506d615184f074752145694999e4bbf1f29220b18d9,1867844866 +RodThroughHoles-v1,level_4,rigid_body/single_arm_dex,5000,0.02,1,50,8,previous_single_task/RodThroughHoles-v1/lora_step_0005000.npz,21e9f0483831cd65a5e75e8415b9b1c496e7ed1714b3b85a1f685184abf4a6ca,1867844866 +SpinPenHora-v1,level_4,rigid_body/single_arm_dex,5000,0.0,0,50,8,previous_single_task/SpinPenHora-v1/lora_step_0005000.npz,e325aba4c998d1b7e3725c8942e13a437e1aee71a5cb8738a2618f31a64e383b,1867844866 +TennisRally-v1,level_4,rigid_body/single_arm_dex,15000,0.06,3,50,8,previous_single_task/TennisRally-v1/lora_step_0015000.npz,2092beb4bec50f1e469c47e456d7589d4b2dc0ae76890845ed06af1de8879b87,1867844866 +FlipPan-v1,level_4,rigid_body/single_arm_gripper,40000,0.04,2,50,8,previous_single_task/FlipPan-v1/lora_step_0040000.npz,4ef6168e12dd59bc8bd8d9dacf8e10df6be69cb341dc4d3c85b2b8c08cc0fa95,1867844866 +HitBilliard-v1,level_4,rigid_body/single_arm_gripper,5000,0.0,0,50,8,previous_single_task/HitBilliard-v1/lora_step_0005000.npz,f047f94f21e69846042a12ec38e75c3bb6e4467f9bfb9eeba39d43db46800748,1867844866 +PingPongBallTransport-v1,level_4,rigid_body/single_arm_gripper,40000,0.16,8,50,7,previous_single_task/PingPongBallTransport-v1/lora_step_0040000.npz,4f62cb281fe6126427b03b9ee60d949ffd7cd6ce38d4b18d7b8203b831d0261a,1867844866 +RingTossing-v1,level_4,rigid_body/single_arm_gripper,5000,0.0,0,50,8,previous_single_task/RingTossing-v1/lora_step_0005000.npz,5b01cd6223af4a4e8cd3a94acd9d6a4c50782845f804a385b4b5719794ede293,1867844866 diff --git a/previous_single_task/manifest.json b/previous_single_task/manifest.json new file mode 100644 index 0000000000000000000000000000000000000000..61bef7e25b42bc02ac13b242efbf44e31156b173 --- /dev/null +++ b/previous_single_task/manifest.json @@ -0,0 +1,815 @@ +{ + "tasks": [ + { + "task": "DualArmCoverCloth-v1", + "level": "level_1", + "suite": "deformable/dual_arm_dex", + "step": 15000, + "success_rate": 0.8, + "successes": 40, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmCoverCloth-v1/lora_step_0015000.npz", + "checkpoint_sha256": "05fe236e482b4a9e6046ac1448bfe952538ed4b7eb217a0961a0e5228f4c91f4", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmRopeStretch-v1", + "level": "level_1", + "suite": "deformable/dual_arm_dex", + "step": 10000, + "success_rate": 0.92, + "successes": 46, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmRopeStretch-v1/lora_step_0010000.npz", + "checkpoint_sha256": "50616181d63062ba47a5a893add02f9b6a99743801d4690f62f11ecb1250feda", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmFoldHandKerchief-v1", + "level": "level_1", + "suite": "deformable/dual_arm_gripper", + "step": 5000, + "success_rate": 0.74, + "successes": 37, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmFoldHandKerchief-v1/lora_step_0005000.npz", + "checkpoint_sha256": "aae641e518457c83346e4ba2c617c3175eb9a18b6d8e3f2c526b20de921868aa", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmPouringWater-v1", + "level": "level_1", + "suite": "deformable/dual_arm_gripper", + "step": 30000, + "success_rate": 0.95, + "successes": 19, + "episodes": 20, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmPouringWater-v1/lora_step_0030000.npz", + "checkpoint_sha256": "d1a0affb3e58bc3eb0b23f5ae83cb9a7f64a8be66940868409a87cb0b1b4ca34", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmRopeCylinderWarpping-v1", + "level": "level_1", + "suite": "deformable/dual_arm_gripper", + "step": 10000, + "success_rate": 0.62, + "successes": 31, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmRopeCylinderWarpping-v1/lora_step_0010000.npz", + "checkpoint_sha256": "a75996da6ae851c12335001a14a54ce38cf3038772fc242de85651ecd4992c6a", + "checkpoint_bytes": 1867844866 + }, + { + "task": "RopePalmCageHoldDex-v1", + "level": "level_1", + "suite": "deformable/single_arm_dex", + "step": 5000, + "success_rate": 0.94, + "successes": 47, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/RopePalmCageHoldDex-v1/lora_step_0005000.npz", + "checkpoint_sha256": "1ce86af8352f14a4738825f6407f5982bb1d7d8c6ece9030bbb98148f39fb05c", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmOpenPen-v1", + "level": "level_1", + "suite": "rigid_body/dual_arm_dex", + "step": 35000, + "success_rate": 0.36, + "successes": 18, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmOpenPen-v1/lora_step_0035000.npz", + "checkpoint_sha256": "8057718dae051ffe7dd5641fc5e50f135068359806d80cd29a273349c008dd78", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmButtonPress-v1", + "level": "level_1", + "suite": "rigid_body/dual_arm_gripper", + "step": 5000, + "success_rate": 0.98, + "successes": 49, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmButtonPress-v1/lora_step_0005000.npz", + "checkpoint_sha256": "fe6b5128698f903f0b7a412f09be569043fdb7f2ae8cb8eb2edc016ff77b1111", + "checkpoint_bytes": 1867844866 + }, + { + "task": "GraspCylinder-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_dex", + "step": 15000, + "success_rate": 0.94, + "successes": 47, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/GraspCylinder-v1/lora_step_0015000.npz", + "checkpoint_sha256": "8df817057bb6ac221d574592a7437a1607aae1853e0350e3426507215748d225", + "checkpoint_bytes": 1867844866 + }, + { + "task": "LiftBucket-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_dex", + "step": 40000, + "success_rate": 0.96, + "successes": 48, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/LiftBucket-v1/lora_step_0040000.npz", + "checkpoint_sha256": "5040809ab7b421527b71dce655b290016467761ca5460a43a32fa2121d2519ee", + "checkpoint_bytes": 1867844866 + }, + { + "task": "TurnOnFaucet-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_dex", + "step": 10000, + "success_rate": 0.92, + "successes": 46, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/TurnOnFaucet-v1/lora_step_0010000.npz", + "checkpoint_sha256": "c233e2a4e9a384ecc107c403938aebb21fc757cc675015d293b96ffd0fce91c0", + "checkpoint_bytes": 1867844866 + }, + { + "task": "OpenDoor-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_gripper", + "step": 5000, + "success_rate": 0.3, + "successes": 15, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/OpenDoor-v1/lora_step_0005000.npz", + "checkpoint_sha256": "9becc4d5effe10eca7baef2875443f2e5ab55c21ec18fbc863537e30280973f7", + "checkpoint_bytes": 1867844866 + }, + { + "task": "PickPlacePen-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_gripper", + "step": 20000, + "success_rate": 0.36, + "successes": 18, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/PickPlacePen-v1/lora_step_0020000.npz", + "checkpoint_sha256": "8969ee8461c5a3ce36cb0a205474c1214990589054aa85ce2f3b1f707e1ab16f", + "checkpoint_bytes": 1867844866 + }, + { + "task": "PushSoccer-v1", + "level": "level_1", + "suite": "rigid_body/single_arm_gripper", + "step": 40000, + "success_rate": 0.8, + "successes": 40, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/PushSoccer-v1/lora_step_0040000.npz", + "checkpoint_sha256": "b495183247ad15943fb3abd608c423928d8ceba2859efef3380fcc4818297a54", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmPlaceRopeOverRod-v1", + "level": "level_2", + "suite": "deformable/dual_arm_dex", + "step": 40000, + "success_rate": 0.46, + "successes": 23, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmPlaceRopeOverRod-v1/lora_step_0040000.npz", + "checkpoint_sha256": "ab110fb81eabcc3bd39bb226cd5856d3b8be169e6ec2745d0b8aebaa94d2ac5e", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmStuffCloth-v1", + "level": "level_2", + "suite": "deformable/dual_arm_dex", + "step": 15000, + "success_rate": 0.4, + "successes": 20, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmStuffCloth-v1/lora_step_0015000.npz", + "checkpoint_sha256": "727b90f7ce8ebdf643ce17e321bf0546117642cc589645f96c2c941e6bb1a7a9", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmClothAlign-v1", + "level": "level_2", + "suite": "deformable/dual_arm_gripper", + "step": 40000, + "success_rate": 0.38, + "successes": 19, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmClothAlign-v1/lora_step_0040000.npz", + "checkpoint_sha256": "c17241b647ad68f11aacc420f950dc5333abd47e7e665e9c0dc84af567943bac", + "checkpoint_bytes": 1867844866 + }, + { + "task": "ClothWipe-v1", + "level": "level_2", + "suite": "deformable/single_arm_dex", + "step": 25000, + "success_rate": 0.52, + "successes": 26, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/ClothWipe-v1/lora_step_0025000.npz", + "checkpoint_sha256": "1603d8b01f721d5ea488ada553dc31fcfad4ac54785d00a639f3d9243e13ae26", + "checkpoint_bytes": 1867844866 + }, + { + "task": "RollingPin-v1", + "level": "level_2", + "suite": "deformable/single_arm_dex", + "step": 25000, + "success_rate": 0.96, + "successes": 48, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/RollingPin-v1/lora_step_0025000.npz", + "checkpoint_sha256": "49c64d6d95014fb02dc5fde1dd3adfac52c28264a1fd6a7a7cd8499a06f410ac", + "checkpoint_bytes": 1867844866 + }, + { + "task": "FoldShirt-v1", + "level": "level_2", + "suite": "deformable/single_arm_gripper", + "step": 40000, + "success_rate": 0.72, + "successes": 36, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/FoldShirt-v1/lora_step_0040000.npz", + "checkpoint_sha256": "594d272a50755c96e1a9cd31a8a961962a4b34c7295a64fc39ed716a4e19cbe0", + "checkpoint_bytes": 1867844866 + }, + { + "task": "StraightenTwoRopesSequential-v1", + "level": "level_2", + "suite": "deformable/single_arm_gripper", + "step": 15000, + "success_rate": 0.36, + "successes": 18, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/StraightenTwoRopesSequential-v1/lora_step_0015000.npz", + "checkpoint_sha256": "9daa2193631b341750a5f0350bdc5e5649018b5fd2af5b2a8783086d4775d9d1", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmOpenBottle-v1", + "level": "level_2", + "suite": "rigid_body/dual_arm_dex", + "step": 30000, + "success_rate": 0.06, + "successes": 3, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmOpenBottle-v1/lora_step_0030000.npz", + "checkpoint_sha256": "badce300065df7e4c45bb4cfb07dfcf74dcd6a33c524f6de76ef3a15aff2490f", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmOpenLaptop-v1", + "level": "level_2", + "suite": "rigid_body/dual_arm_dex", + "step": 25000, + "success_rate": 0.42, + "successes": 21, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmOpenLaptop-v1/lora_step_0025000.npz", + "checkpoint_sha256": "96e016a4d0628be868f5ae1108397addc4216fb3123020171775346ad335cca3", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmThrowBlock-v1", + "level": "level_2", + "suite": "rigid_body/dual_arm_dex", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmThrowBlock-v1/lora_step_0005000.npz", + "checkpoint_sha256": "ddc085746228651a8ef69d51c548e10e6c50ca76081808df23890913a3bf20e0", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmPegInHole-v1", + "level": "level_2", + "suite": "rigid_body/dual_arm_gripper", + "step": 5000, + "success_rate": 0.18, + "successes": 9, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmPegInHole-v1/lora_step_0005000.npz", + "checkpoint_sha256": "3da1d0202466bf4a0182a47ddb00ecd42112a22ed462c85197a9b5625f4d5508", + "checkpoint_bytes": 1867844866 + }, + { + "task": "GraspTwoBlocks-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_dex", + "step": 5000, + "success_rate": 0.72, + "successes": 36, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/GraspTwoBlocks-v1/lora_step_0005000.npz", + "checkpoint_sha256": "84de21676c4ded27f090d19e152584578a07ce0de6816d732e84f967a9378831", + "checkpoint_bytes": 1867844866 + }, + { + "task": "HammerNailOut-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_dex", + "step": 40000, + "success_rate": 0.44, + "successes": 22, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/HammerNailOut-v1/lora_step_0040000.npz", + "checkpoint_sha256": "8dc1babeb6ad28356e7a4fedc58122db41c17d4564e0ac1d95cd7c8fbe5b228a", + "checkpoint_bytes": 1867844866 + }, + { + "task": "ReorientPen-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_dex", + "step": 25000, + "success_rate": 0.38, + "successes": 19, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/ReorientPen-v1/lora_step_0025000.npz", + "checkpoint_sha256": "070504c2d43469bc87fd254b20ab68c2c4b646c9acf7fc066a6a5ad0e5bde174", + "checkpoint_bytes": 1867844866 + }, + { + "task": "HangMug-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_gripper", + "step": 10000, + "success_rate": 0.7, + "successes": 35, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/HangMug-v1/lora_step_0010000.npz", + "checkpoint_sha256": "7ad647a1261390f68a70fa7320c3238a05aec39db9d37d8a34be76c63ee1c8ed", + "checkpoint_bytes": 1867844866 + }, + { + "task": "KitAssembly-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_gripper", + "step": 15000, + "success_rate": 0.06, + "successes": 3, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/KitAssembly-v1/lora_step_0015000.npz", + "checkpoint_sha256": "2e78e125fcb2aacf7a1c59973868de4839dc8bf7c6fdfa649b5ebde9b3841189", + "checkpoint_bytes": 1867844866 + }, + { + "task": "PivotBook-v1", + "level": "level_2", + "suite": "rigid_body/single_arm_gripper", + "step": 15000, + "success_rate": 0.38, + "successes": 19, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/PivotBook-v1/lora_step_0015000.npz", + "checkpoint_sha256": "1f419832345ac22ea2f4a05686c32c7670f62d2704c5b5025e19414cfa51b047", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmShapeDoll-v1", + "level": "level_3", + "suite": "deformable/dual_arm_gripper", + "step": 30000, + "success_rate": 0.04, + "successes": 2, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmShapeDoll-v1/lora_step_0030000.npz", + "checkpoint_sha256": "e290f251fb1feb05077deb5167416a0bf34c273229cff2ad60b0dfbd91934955", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmThreadRopeThroughTube-v1", + "level": "level_3", + "suite": "deformable/dual_arm_gripper", + "step": 35000, + "success_rate": 0.16, + "successes": 8, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmThreadRopeThroughTube-v1/lora_step_0035000.npz", + "checkpoint_sha256": "ed640c43be58457fab4969a591ef4ab14103d5553c3f363f39412e0ff47bbdd1", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmWeaveCloth-v1", + "level": "level_3", + "suite": "deformable/dual_arm_gripper", + "step": 20000, + "success_rate": 0.32, + "successes": 16, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmWeaveCloth-v1/lora_step_0020000.npz", + "checkpoint_sha256": "0db96e943ccf5a38c41f132f289033e36e87358deb637445b03c9fdc37a1bd60", + "checkpoint_bytes": 1867844866 + }, + { + "task": "HitBox-v1", + "level": "level_3", + "suite": "deformable/single_arm_dex", + "step": 25000, + "success_rate": 0.06, + "successes": 3, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/HitBox-v1/lora_step_0025000.npz", + "checkpoint_sha256": "fc4b6cdf6398e6b7222b8a9ba80630fbd262fb2de923829c55aa3aa1c2df27c0", + "checkpoint_bytes": 1867844866 + }, + { + "task": "LiquidMeasuring-v1", + "level": "level_3", + "suite": "deformable/single_arm_gripper", + "step": 35000, + "success_rate": 0.32, + "successes": 16, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/LiquidMeasuring-v1/lora_step_0035000.npz", + "checkpoint_sha256": "6e0a9a12b8f5882b391f082b0640baae09dcab280030c3e437804a56f1fa7a8f", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmPingpongBalance-v1", + "level": "level_3", + "suite": "rigid_body/dual_arm_dex", + "step": 15000, + "success_rate": 0.2, + "successes": 10, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmPingpongBalance-v1/lora_step_0015000.npz", + "checkpoint_sha256": "da1fb459a8bcc79cb0985cb6e78eb4e477f7c2a88908422063379612734dab5d", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmSetupTable-v1", + "level": "level_3", + "suite": "rigid_body/dual_arm_dex", + "step": 25000, + "success_rate": 0.04, + "successes": 2, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmSetupTable-v1/lora_step_0025000.npz", + "checkpoint_sha256": "093e094f8ea395f959f4160b04c57930e33c5a3df4cc6504428a877b03a2244c", + "checkpoint_bytes": 1867992394 + }, + { + "task": "DualArmOpenFridge-v1", + "level": "level_3", + "suite": "rigid_body/dual_arm_gripper", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmOpenFridge-v1/lora_step_0005000.npz", + "checkpoint_sha256": "311033395a4c57babcae115e5a0ae2c179504482d1d85075956fd03d545c5e48", + "checkpoint_bytes": 1867844866 + }, + { + "task": "HammerNail-v1", + "level": "level_3", + "suite": "rigid_body/single_arm_dex", + "step": 35000, + "success_rate": 0.12, + "successes": 6, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/HammerNail-v1/lora_step_0035000.npz", + "checkpoint_sha256": "a26699d0543203305445941f7008a0c0ed01afa2b451ffa3850d54fbc45f4dc9", + "checkpoint_bytes": 1867844866 + }, + { + "task": "MobilePickPlace-v1", + "level": "level_3", + "suite": "rigid_body/single_arm_dex", + "step": 15000, + "success_rate": 0.08, + "successes": 4, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/MobilePickPlace-v1/lora_step_0015000.npz", + "checkpoint_sha256": "a5f73d09c256f142d64b4edf03062bc328f6a3d0c13c85a0949a04d875cd64cb", + "checkpoint_bytes": 1867844866 + }, + { + "task": "TweezersPickPlace-v1", + "level": "level_3", + "suite": "rigid_body/single_arm_dex", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/TweezersPickPlace-v1/lora_step_0005000.npz", + "checkpoint_sha256": "23135a349b12ab66e78abcd43364a7f1c0f6a05b13ae08bc8cfb626005f6d306", + "checkpoint_bytes": 1867844866 + }, + { + "task": "MazeBlockPush-v1", + "level": "level_3", + "suite": "rigid_body/single_arm_gripper", + "step": 30000, + "success_rate": 0.65, + "successes": 13, + "episodes": 20, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/MazeBlockPush-v1/lora_step_0030000.npz", + "checkpoint_sha256": "e7396df9ba9f96b5083f9ca0b1e128b7ce0bc88b8eb5319987eb033f60269602", + "checkpoint_bytes": 1867844866 + }, + { + "task": "Stack6Blocks-v1", + "level": "level_3", + "suite": "rigid_body/single_arm_gripper", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/Stack6Blocks-v1/lora_step_0005000.npz", + "checkpoint_sha256": "9ee601932f1d2357e3007185e9d991bf04d45e59123ba76950d9a70041c9c300", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmJumpRope-v1", + "level": "level_4", + "suite": "deformable/dual_arm_dex", + "step": 20000, + "success_rate": 0.02, + "successes": 1, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmJumpRope-v1/lora_step_0020000.npz", + "checkpoint_sha256": "15b86fb5ec34e19d2161c8f535d0ba3c2d0917f4b1434d574d1bb56ee7367487", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmClothHanger-v1", + "level": "level_4", + "suite": "deformable/dual_arm_gripper", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmClothHanger-v1/lora_step_0005000.npz", + "checkpoint_sha256": "7c5dae817921bdcdf2474ea67dc5e7edd59fb41832db3e06af5ffc9f1f60c28c", + "checkpoint_bytes": 1867844866 + }, + { + "task": "BuildTent-v1", + "level": "level_4", + "suite": "deformable/single_arm_gripper", + "step": 10000, + "success_rate": 0.06, + "successes": 3, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/BuildTent-v1/lora_step_0010000.npz", + "checkpoint_sha256": "d27916914df184fddd0df81bde6d7f7187f76d203648989b74d2711fca67281d", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmFlipTable-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_dex", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmFlipTable-v1/lora_step_0005000.npz", + "checkpoint_sha256": "84f91077bad9c28bf3fd78e16d44e41b3de8a988f36eea6b6aa3346e44f7790a", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmKeyboardType-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_dex", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 7, + "checkpoint": "previous_single_task/DualArmKeyboardType-v1/lora_step_0005000.npz", + "checkpoint_sha256": "521d586261c29c45cf239aebd59190737d7be7d05cc1afd943f361e72fa76bee", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmSendPingpong-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_dex", + "step": 40000, + "success_rate": 0.08, + "successes": 4, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmSendPingpong-v1/lora_step_0040000.npz", + "checkpoint_sha256": "70924c6c7880fc34d61197d289cbbbbdc15a3e146339aed6b10af36f19ea3126", + "checkpoint_bytes": 1867943218 + }, + { + "task": "DualArmShootBasketball-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_dex", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmShootBasketball-v1/lora_step_0005000.npz", + "checkpoint_sha256": "03ce4d67db10529e152c37e5894df057c430f38592273d6ec47cbbd4500226e7", + "checkpoint_bytes": 1867992394 + }, + { + "task": "DualArmBallClassify-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_gripper", + "step": 10000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 7, + "checkpoint": "previous_single_task/DualArmBallClassify-v1/lora_step_0010000.npz", + "checkpoint_sha256": "0bec5644e89c77f1fbdb520eae5ea4504903d0bbed7dab3b41295d045e50aa0f", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmBilliardBridge-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_gripper", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmBilliardBridge-v1/lora_step_0005000.npz", + "checkpoint_sha256": "7dabff95c821665a5f6ad6d381ca64ce8dc8ae1baa11aa6b1cc042d3ebc6bc9e", + "checkpoint_bytes": 1867844866 + }, + { + "task": "DualArmMahjongSetup-v1", + "level": "level_4", + "suite": "rigid_body/dual_arm_gripper", + "step": 5000, + "success_rate": 0.02, + "successes": 1, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/DualArmMahjongSetup-v1/lora_step_0005000.npz", + "checkpoint_sha256": "f565771d810c4e9c7d1499feee6ffbe540235cabab920b80634365b74c11a51a", + "checkpoint_bytes": 1867844866 + }, + { + "task": "FitnessBall-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_dex", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/FitnessBall-v1/lora_step_0005000.npz", + "checkpoint_sha256": "063779d861fa3fb188065506d615184f074752145694999e4bbf1f29220b18d9", + "checkpoint_bytes": 1867844866 + }, + { + "task": "RodThroughHoles-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_dex", + "step": 5000, + "success_rate": 0.02, + "successes": 1, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/RodThroughHoles-v1/lora_step_0005000.npz", + "checkpoint_sha256": "21e9f0483831cd65a5e75e8415b9b1c496e7ed1714b3b85a1f685184abf4a6ca", + "checkpoint_bytes": 1867844866 + }, + { + "task": "SpinPenHora-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_dex", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/SpinPenHora-v1/lora_step_0005000.npz", + "checkpoint_sha256": "e325aba4c998d1b7e3725c8942e13a437e1aee71a5cb8738a2618f31a64e383b", + "checkpoint_bytes": 1867844866 + }, + { + "task": "TennisRally-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_dex", + "step": 15000, + "success_rate": 0.06, + "successes": 3, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/TennisRally-v1/lora_step_0015000.npz", + "checkpoint_sha256": "2092beb4bec50f1e469c47e456d7589d4b2dc0ae76890845ed06af1de8879b87", + "checkpoint_bytes": 1867844866 + }, + { + "task": "FlipPan-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_gripper", + "step": 40000, + "success_rate": 0.04, + "successes": 2, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/FlipPan-v1/lora_step_0040000.npz", + "checkpoint_sha256": "4ef6168e12dd59bc8bd8d9dacf8e10df6be69cb341dc4d3c85b2b8c08cc0fa95", + "checkpoint_bytes": 1867844866 + }, + { + "task": "HitBilliard-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_gripper", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/HitBilliard-v1/lora_step_0005000.npz", + "checkpoint_sha256": "f047f94f21e69846042a12ec38e75c3bb6e4467f9bfb9eeba39d43db46800748", + "checkpoint_bytes": 1867844866 + }, + { + "task": "PingPongBallTransport-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_gripper", + "step": 40000, + "success_rate": 0.16, + "successes": 8, + "episodes": 50, + "ckpts_evaluated": 7, + "checkpoint": "previous_single_task/PingPongBallTransport-v1/lora_step_0040000.npz", + "checkpoint_sha256": "4f62cb281fe6126427b03b9ee60d949ffd7cd6ce38d4b18d7b8203b831d0261a", + "checkpoint_bytes": 1867844866 + }, + { + "task": "RingTossing-v1", + "level": "level_4", + "suite": "rigid_body/single_arm_gripper", + "step": 5000, + "success_rate": 0.0, + "successes": 0, + "episodes": 50, + "ckpts_evaluated": 8, + "checkpoint": "previous_single_task/RingTossing-v1/lora_step_0005000.npz", + "checkpoint_sha256": "5b01cd6223af4a4e8cd3a94acd9d6a4c50782845f804a385b4b5719794ede293", + "checkpoint_bytes": 1867844866 + } + ], + "selection": "one best checkpoint from each of the 62 previously reusable single-task runs", + "eval50_tasks": 60, + "eval20_tasks": 2, + "audit_date": "2026-10-22", + "note": "Historical collection; newer retraining runs under single_task/ are preserved. Episode counts are explicit and must not be mixed." +}