Add PyRUA-Dex evaluation logs

#7
by VME50121 - opened
This view is limited to 50 files because it contains too many changes. See the raw diff here.
Files changed (50) hide show
  1. pyrua-dex/README.md +28 -0
  2. pyrua-dex/leaderboard_pyrua-dex.json +2446 -0
  3. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/native_shards/000/per_episode.jsonl +0 -0
  4. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/native_shards/000/summary.json +193 -0
  5. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/native_shards/001/per_episode.jsonl +0 -0
  6. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/native_shards/001/summary.json +187 -0
  7. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/per_episode.jsonl +0 -0
  8. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/summary.json +195 -0
  9. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/native_shards/000/per_episode.jsonl +0 -0
  10. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/native_shards/000/summary.json +185 -0
  11. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/native_shards/001/per_episode.jsonl +0 -0
  12. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/native_shards/001/summary.json +187 -0
  13. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/per_episode.jsonl +0 -0
  14. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/summary.json +188 -0
  15. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/native_shards/000/per_episode.jsonl +0 -0
  16. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/native_shards/000/summary.json +187 -0
  17. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/native_shards/001/per_episode.jsonl +0 -0
  18. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/native_shards/001/summary.json +193 -0
  19. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/per_episode.jsonl +0 -0
  20. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/summary.json +195 -0
  21. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/native_shards/000/per_episode.jsonl +0 -0
  22. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/native_shards/000/summary.json +185 -0
  23. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/native_shards/001/per_episode.jsonl +0 -0
  24. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/native_shards/001/summary.json +189 -0
  25. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/per_episode.jsonl +0 -0
  26. pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/summary.json +189 -0
  27. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/native_shards/000/per_episode.jsonl +0 -0
  28. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/native_shards/000/summary.json +222 -0
  29. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/native_shards/001/per_episode.jsonl +0 -0
  30. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/native_shards/001/summary.json +221 -0
  31. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/per_episode.jsonl +0 -0
  32. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/summary.json +231 -0
  33. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/native_shards/000/per_episode.jsonl +0 -0
  34. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/native_shards/000/summary.json +227 -0
  35. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/native_shards/001/per_episode.jsonl +0 -0
  36. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/native_shards/001/summary.json +213 -0
  37. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/per_episode.jsonl +0 -0
  38. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/summary.json +229 -0
  39. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/native_shards/000/per_episode.jsonl +0 -0
  40. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/native_shards/000/summary.json +214 -0
  41. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/native_shards/001/per_episode.jsonl +0 -0
  42. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/native_shards/001/summary.json +216 -0
  43. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/per_episode.jsonl +0 -0
  44. pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/summary.json +221 -0
  45. pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/native_shards/000/per_episode.jsonl +0 -0
  46. pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/native_shards/000/summary.json +225 -0
  47. pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/native_shards/001/per_episode.jsonl +0 -0
  48. pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/native_shards/001/summary.json +212 -0
  49. pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/per_episode.jsonl +0 -0
  50. pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/summary.json +226 -0
pyrua-dex/README.md ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # PyRUA-Dex
2
+
3
+ **PKU DAGroup**
4
+
5
+ PyRUA-Dex is a coding-agent system for dexterous manipulation. **No human demonstration data is used.** For each
6
+ task, coding agents (GPT-6 Astra) explore in simulation and write a controller program; the best program is frozen and
7
+ evaluated alone, with no trained weights and no language-model calls during evaluation.
8
+
9
+ [Project](https://vme0522.github.io/PyRUA-Dex/) · [Code](https://github.com/DAGroup-PKU/PyRUA-Lean/tree/dex) (coming soon) · arXiv: coming soon
10
+
11
+ **Inputs:** Vision + Proprioception + Tactile
12
+
13
+ ## Evaluation
14
+
15
+ 26 tasks · None / Equi. / Inv. / Full · 50 episodes per task/channel (5,200 episodes), with the official seeds and episode limits.
16
+
17
+ | Metric | Result |
18
+ | --- | ---: |
19
+ | None SR | 44.9% |
20
+ | Equi. SR | 45.8% |
21
+ | Inv. SR | 44.3% |
22
+ | Full SR | 43.4% |
23
+ | Four-channel mean SR | 44.6% |
24
+ | Four-channel mean LSCR | 67.0% |
25
+ | Efficiency (22 tasks with success) | 0.889 |
26
+
27
+ Protocol differences are listed in `evaluation.protocol.protocolChanges` of the results JSON.
28
+ [Evaluation logs](https://huggingface.co/datasets/Bench2Dex/Leaderboard/tree/main/pyrua-dex).
pyrua-dex/leaderboard_pyrua-dex.json ADDED
@@ -0,0 +1,2446 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "model": {
3
+ "id": "pyrua-dex",
4
+ "label": "PyRUA-Dex",
5
+ "variantLabel": "",
6
+ "family": "PyRUA",
7
+ "modelVersion": "PyRUA-Dex round 1 (coding agents: GPT-6 Astra)",
8
+ "checkpoint": null,
9
+ "inputs": "RGB cameras + proprioception + tactile + camera calibration + task description with goal text",
10
+ "protocolId": "standard",
11
+ "displayOrder": 130,
12
+ "trainingSetting": "No human demonstration data and no weight updates: coding agents write and test controller programs on development layouts (seeds disjoint from the evaluation seeds); the best program per task is frozen",
13
+ "resultVersion": "Round 1 (2026-11-15)",
14
+ "updatedAt": "2026-11-16",
15
+ "evaluationSetting": "One frozen program per task; 26 tasks x 4 channels x 50 episodes; no language-model calls during evaluation",
16
+ "sourceLabel": "PyRUA-Dex project",
17
+ "sourceUrl": "https://vme0522.github.io/PyRUA-Dex/",
18
+ "resultDetailsUrl": "https://huggingface.co/datasets/Bench2Dex/Leaderboard/tree/main/pyrua-dex",
19
+ "contributor": "PKU DAGroup",
20
+ "contributorUrl": "https://github.com/DAGroup-PKU",
21
+ "type": "Agent",
22
+ "modelUrl": "https://vme0522.github.io/PyRUA-Dex/",
23
+ "logoPath": "leaderboard/pyrua-dex/logo.svg",
24
+ "logoAlt": "Peking University",
25
+ "weightsUrl": null,
26
+ "codeUrl": "https://github.com/DAGroup-PKU/PyRUA-Lean/tree/dex"
27
+ },
28
+ "overall": {
29
+ "taskCount": 26,
30
+ "scope": "task-macro",
31
+ "sr": 44.61538461538461,
32
+ "lscr": 67.02115384615385,
33
+ "efficiency": 0.8889243062360126,
34
+ "efficiencyStatus": "reported",
35
+ "efficiencyTaskCount": 22,
36
+ "channelSr": {
37
+ "None": 44.92307692307692,
38
+ "Equi.": 45.84615384615385,
39
+ "Inv.": 44.30769230769231,
40
+ "Full": 43.38461538461539
41
+ },
42
+ "channelLscr": {
43
+ "None": 67.6525641025641,
44
+ "Equi.": 67.85128205128206,
45
+ "Inv.": 66.68589743589743,
46
+ "Full": 65.89487179487179
47
+ },
48
+ "channelTaskCounts": {
49
+ "None": 26,
50
+ "Equi.": 26,
51
+ "Inv.": 26,
52
+ "Full": 26
53
+ },
54
+ "status": "reported"
55
+ },
56
+ "per_task": {
57
+ "03": {
58
+ "channels": {
59
+ "None": {
60
+ "status": "reported",
61
+ "episodes": 50,
62
+ "successes": 39,
63
+ "lscr": 94.66666666666667,
64
+ "source": {
65
+ "runId": "pyrua-dex_r01_03_wine_fast-e9124b6af8",
66
+ "perEpisodeFile": "per_task_eval/03_wine_glass_plate_balance/None/per_episode.jsonl",
67
+ "summaryFile": "per_task_eval/03_wine_glass_plate_balance/None/summary.json"
68
+ }
69
+ },
70
+ "Equi.": {
71
+ "status": "reported",
72
+ "episodes": 50,
73
+ "successes": 38,
74
+ "lscr": 90.66666666666666,
75
+ "source": {
76
+ "runId": "pyrua-dex_r01_03_wine_fast-e9124b6af8",
77
+ "perEpisodeFile": "per_task_eval/03_wine_glass_plate_balance/Equi/per_episode.jsonl",
78
+ "summaryFile": "per_task_eval/03_wine_glass_plate_balance/Equi/summary.json"
79
+ }
80
+ },
81
+ "Inv.": {
82
+ "status": "reported",
83
+ "episodes": 50,
84
+ "successes": 37,
85
+ "lscr": 92.0,
86
+ "source": {
87
+ "runId": "pyrua-dex_r01_03_wine_fast-e9124b6af8",
88
+ "perEpisodeFile": "per_task_eval/03_wine_glass_plate_balance/Inv/per_episode.jsonl",
89
+ "summaryFile": "per_task_eval/03_wine_glass_plate_balance/Inv/summary.json"
90
+ }
91
+ },
92
+ "Full": {
93
+ "status": "reported",
94
+ "episodes": 50,
95
+ "successes": 40,
96
+ "lscr": 92.66666666666666,
97
+ "source": {
98
+ "runId": "pyrua-dex_r01_03_wine_fast-e9124b6af8",
99
+ "perEpisodeFile": "per_task_eval/03_wine_glass_plate_balance/Full/per_episode.jsonl",
100
+ "summaryFile": "per_task_eval/03_wine_glass_plate_balance/Full/summary.json"
101
+ }
102
+ }
103
+ },
104
+ "mean": {
105
+ "status": "reported",
106
+ "sr": 77.0,
107
+ "srScope": "four-channel-mean",
108
+ "srEpisodes": 200,
109
+ "lscr": 92.5,
110
+ "lscrScope": "four-channel-mean",
111
+ "lscrEpisodes": 200,
112
+ "efficiency": 1.0444244564859706,
113
+ "efficiencyStatus": "reported"
114
+ }
115
+ },
116
+ "06": {
117
+ "channels": {
118
+ "None": {
119
+ "status": "reported",
120
+ "episodes": 50,
121
+ "successes": 26,
122
+ "lscr": 87.0,
123
+ "source": {
124
+ "runId": "pyrua-dex_r01_06_fruit_controller-429ffea065",
125
+ "perEpisodeFile": "per_task_eval/06_fruit_bowl_loading/None/per_episode.jsonl",
126
+ "summaryFile": "per_task_eval/06_fruit_bowl_loading/None/summary.json"
127
+ }
128
+ },
129
+ "Equi.": {
130
+ "status": "reported",
131
+ "episodes": 50,
132
+ "successes": 29,
133
+ "lscr": 80.5,
134
+ "source": {
135
+ "runId": "pyrua-dex_r01_06_fruit_controller-429ffea065",
136
+ "perEpisodeFile": "per_task_eval/06_fruit_bowl_loading/Equi/per_episode.jsonl",
137
+ "summaryFile": "per_task_eval/06_fruit_bowl_loading/Equi/summary.json"
138
+ }
139
+ },
140
+ "Inv.": {
141
+ "status": "reported",
142
+ "episodes": 50,
143
+ "successes": 21,
144
+ "lscr": 78.5,
145
+ "source": {
146
+ "runId": "pyrua-dex_r01_06_fruit_controller-429ffea065",
147
+ "perEpisodeFile": "per_task_eval/06_fruit_bowl_loading/Inv/per_episode.jsonl",
148
+ "summaryFile": "per_task_eval/06_fruit_bowl_loading/Inv/summary.json"
149
+ }
150
+ },
151
+ "Full": {
152
+ "status": "reported",
153
+ "episodes": 50,
154
+ "successes": 26,
155
+ "lscr": 79.5,
156
+ "source": {
157
+ "runId": "pyrua-dex_r01_06_fruit_controller-429ffea065",
158
+ "perEpisodeFile": "per_task_eval/06_fruit_bowl_loading/Full/per_episode.jsonl",
159
+ "summaryFile": "per_task_eval/06_fruit_bowl_loading/Full/summary.json"
160
+ }
161
+ }
162
+ },
163
+ "mean": {
164
+ "status": "reported",
165
+ "sr": 51.0,
166
+ "srScope": "four-channel-mean",
167
+ "srEpisodes": 200,
168
+ "lscr": 81.375,
169
+ "lscrScope": "four-channel-mean",
170
+ "lscrEpisodes": 200,
171
+ "efficiency": 0.7053789466859723,
172
+ "efficiencyStatus": "reported"
173
+ }
174
+ },
175
+ "07": {
176
+ "channels": {
177
+ "None": {
178
+ "status": "reported",
179
+ "episodes": 50,
180
+ "successes": 35,
181
+ "lscr": 82.5,
182
+ "source": {
183
+ "runId": "pyrua-dex_r01_07_citrus_policy-3695867c40",
184
+ "perEpisodeFile": "per_task_eval/07_citrus_plate_loading/None/per_episode.jsonl",
185
+ "summaryFile": "per_task_eval/07_citrus_plate_loading/None/summary.json"
186
+ }
187
+ },
188
+ "Equi.": {
189
+ "status": "reported",
190
+ "episodes": 50,
191
+ "successes": 41,
192
+ "lscr": 90.0,
193
+ "source": {
194
+ "runId": "pyrua-dex_r01_07_citrus_policy-3695867c40",
195
+ "perEpisodeFile": "per_task_eval/07_citrus_plate_loading/Equi/per_episode.jsonl",
196
+ "summaryFile": "per_task_eval/07_citrus_plate_loading/Equi/summary.json"
197
+ }
198
+ },
199
+ "Inv.": {
200
+ "status": "reported",
201
+ "episodes": 50,
202
+ "successes": 38,
203
+ "lscr": 81.5,
204
+ "source": {
205
+ "runId": "pyrua-dex_r01_07_citrus_policy-3695867c40",
206
+ "perEpisodeFile": "per_task_eval/07_citrus_plate_loading/Inv/per_episode.jsonl",
207
+ "summaryFile": "per_task_eval/07_citrus_plate_loading/Inv/summary.json"
208
+ }
209
+ },
210
+ "Full": {
211
+ "status": "reported",
212
+ "episodes": 50,
213
+ "successes": 41,
214
+ "lscr": 90.5,
215
+ "source": {
216
+ "runId": "pyrua-dex_r01_07_citrus_policy-3695867c40",
217
+ "perEpisodeFile": "per_task_eval/07_citrus_plate_loading/Full/per_episode.jsonl",
218
+ "summaryFile": "per_task_eval/07_citrus_plate_loading/Full/summary.json"
219
+ }
220
+ }
221
+ },
222
+ "mean": {
223
+ "status": "reported",
224
+ "sr": 77.5,
225
+ "srScope": "four-channel-mean",
226
+ "srEpisodes": 200,
227
+ "lscr": 86.125,
228
+ "lscrScope": "four-channel-mean",
229
+ "lscrEpisodes": 200,
230
+ "efficiency": 0.8214063883492579,
231
+ "efficiencyStatus": "reported"
232
+ }
233
+ },
234
+ "08": {
235
+ "channels": {
236
+ "None": {
237
+ "status": "reported",
238
+ "episodes": 50,
239
+ "successes": 0,
240
+ "lscr": 51.0,
241
+ "source": {
242
+ "runId": "pyrua-dex_r01_08_soy_fixed-cdfd94bffc",
243
+ "perEpisodeFile": "per_task_eval/08_frypan_stand_pour/None/per_episode.jsonl",
244
+ "summaryFile": "per_task_eval/08_frypan_stand_pour/None/summary.json"
245
+ }
246
+ },
247
+ "Equi.": {
248
+ "status": "reported",
249
+ "episodes": 50,
250
+ "successes": 0,
251
+ "lscr": 51.0,
252
+ "source": {
253
+ "runId": "pyrua-dex_r01_08_soy_fixed-cdfd94bffc",
254
+ "perEpisodeFile": "per_task_eval/08_frypan_stand_pour/Equi/per_episode.jsonl",
255
+ "summaryFile": "per_task_eval/08_frypan_stand_pour/Equi/summary.json"
256
+ }
257
+ },
258
+ "Inv.": {
259
+ "status": "reported",
260
+ "episodes": 50,
261
+ "successes": 0,
262
+ "lscr": 43.0,
263
+ "source": {
264
+ "runId": "pyrua-dex_r01_08_soy_fixed-cdfd94bffc",
265
+ "perEpisodeFile": "per_task_eval/08_frypan_stand_pour/Inv/per_episode.jsonl",
266
+ "summaryFile": "per_task_eval/08_frypan_stand_pour/Inv/summary.json"
267
+ }
268
+ },
269
+ "Full": {
270
+ "status": "reported",
271
+ "episodes": 50,
272
+ "successes": 0,
273
+ "lscr": 40.0,
274
+ "source": {
275
+ "runId": "pyrua-dex_r01_08_soy_fixed-cdfd94bffc",
276
+ "perEpisodeFile": "per_task_eval/08_frypan_stand_pour/Full/per_episode.jsonl",
277
+ "summaryFile": "per_task_eval/08_frypan_stand_pour/Full/summary.json"
278
+ }
279
+ }
280
+ },
281
+ "mean": {
282
+ "status": "reported",
283
+ "sr": 0.0,
284
+ "srScope": "four-channel-mean",
285
+ "srEpisodes": 200,
286
+ "lscr": 46.25,
287
+ "lscrScope": "four-channel-mean",
288
+ "lscrEpisodes": 200,
289
+ "efficiency": null,
290
+ "efficiencyStatus": "no-success"
291
+ }
292
+ },
293
+ "09": {
294
+ "channels": {
295
+ "None": {
296
+ "status": "reported",
297
+ "episodes": 50,
298
+ "successes": 41,
299
+ "lscr": 88.0,
300
+ "source": {
301
+ "runId": "pyrua-dex_r01_09_cleaner55-5f88a39d99",
302
+ "perEpisodeFile": "per_task_eval/09_cleaner_moisturizer_box_loading/None/per_episode.jsonl",
303
+ "summaryFile": "per_task_eval/09_cleaner_moisturizer_box_loading/None/summary.json"
304
+ }
305
+ },
306
+ "Equi.": {
307
+ "status": "reported",
308
+ "episodes": 50,
309
+ "successes": 33,
310
+ "lscr": 80.0,
311
+ "source": {
312
+ "runId": "pyrua-dex_r01_09_cleaner55-5f88a39d99",
313
+ "perEpisodeFile": "per_task_eval/09_cleaner_moisturizer_box_loading/Equi/per_episode.jsonl",
314
+ "summaryFile": "per_task_eval/09_cleaner_moisturizer_box_loading/Equi/summary.json"
315
+ }
316
+ },
317
+ "Inv.": {
318
+ "status": "reported",
319
+ "episodes": 50,
320
+ "successes": 35,
321
+ "lscr": 81.0,
322
+ "source": {
323
+ "runId": "pyrua-dex_r01_09_cleaner55-5f88a39d99",
324
+ "perEpisodeFile": "per_task_eval/09_cleaner_moisturizer_box_loading/Inv/per_episode.jsonl",
325
+ "summaryFile": "per_task_eval/09_cleaner_moisturizer_box_loading/Inv/summary.json"
326
+ }
327
+ },
328
+ "Full": {
329
+ "status": "reported",
330
+ "episodes": 50,
331
+ "successes": 36,
332
+ "lscr": 83.0,
333
+ "source": {
334
+ "runId": "pyrua-dex_r01_09_cleaner55-5f88a39d99",
335
+ "perEpisodeFile": "per_task_eval/09_cleaner_moisturizer_box_loading/Full/per_episode.jsonl",
336
+ "summaryFile": "per_task_eval/09_cleaner_moisturizer_box_loading/Full/summary.json"
337
+ }
338
+ }
339
+ },
340
+ "mean": {
341
+ "status": "reported",
342
+ "sr": 72.5,
343
+ "srScope": "four-channel-mean",
344
+ "srEpisodes": 200,
345
+ "lscr": 83.0,
346
+ "lscrScope": "four-channel-mean",
347
+ "lscrEpisodes": 200,
348
+ "efficiency": 0.8321740258462199,
349
+ "efficiencyStatus": "reported"
350
+ }
351
+ },
352
+ "12": {
353
+ "channels": {
354
+ "None": {
355
+ "status": "reported",
356
+ "episodes": 50,
357
+ "successes": 0,
358
+ "lscr": 52.5,
359
+ "source": {
360
+ "runId": "pyrua-dex_r01_12_measured_tools-adce3bd7ad",
361
+ "perEpisodeFile": "per_task_eval/12_screwdriver_box_and_hammer/None/per_episode.jsonl",
362
+ "summaryFile": "per_task_eval/12_screwdriver_box_and_hammer/None/summary.json"
363
+ }
364
+ },
365
+ "Equi.": {
366
+ "status": "reported",
367
+ "episodes": 50,
368
+ "successes": 0,
369
+ "lscr": 53.5,
370
+ "source": {
371
+ "runId": "pyrua-dex_r01_12_measured_tools-adce3bd7ad",
372
+ "perEpisodeFile": "per_task_eval/12_screwdriver_box_and_hammer/Equi/per_episode.jsonl",
373
+ "summaryFile": "per_task_eval/12_screwdriver_box_and_hammer/Equi/summary.json"
374
+ }
375
+ },
376
+ "Inv.": {
377
+ "status": "reported",
378
+ "episodes": 50,
379
+ "successes": 0,
380
+ "lscr": 53.0,
381
+ "source": {
382
+ "runId": "pyrua-dex_r01_12_measured_tools-adce3bd7ad",
383
+ "perEpisodeFile": "per_task_eval/12_screwdriver_box_and_hammer/Inv/per_episode.jsonl",
384
+ "summaryFile": "per_task_eval/12_screwdriver_box_and_hammer/Inv/summary.json"
385
+ }
386
+ },
387
+ "Full": {
388
+ "status": "reported",
389
+ "episodes": 50,
390
+ "successes": 0,
391
+ "lscr": 55.50000000000001,
392
+ "source": {
393
+ "runId": "pyrua-dex_r01_12_measured_tools-adce3bd7ad",
394
+ "perEpisodeFile": "per_task_eval/12_screwdriver_box_and_hammer/Full/per_episode.jsonl",
395
+ "summaryFile": "per_task_eval/12_screwdriver_box_and_hammer/Full/summary.json"
396
+ }
397
+ }
398
+ },
399
+ "mean": {
400
+ "status": "reported",
401
+ "sr": 0.0,
402
+ "srScope": "four-channel-mean",
403
+ "srEpisodes": 200,
404
+ "lscr": 53.625,
405
+ "lscrScope": "four-channel-mean",
406
+ "lscrEpisodes": 200,
407
+ "efficiency": null,
408
+ "efficiencyStatus": "no-success"
409
+ }
410
+ },
411
+ "21": {
412
+ "channels": {
413
+ "None": {
414
+ "status": "reported",
415
+ "episodes": 50,
416
+ "successes": 8,
417
+ "lscr": 45.33333333333333,
418
+ "source": {
419
+ "runId": "pyrua-dex_r01_21_round_fallback-33066e1d92",
420
+ "perEpisodeFile": "per_task_eval/21_condiment_box_loading/None/per_episode.jsonl",
421
+ "summaryFile": "per_task_eval/21_condiment_box_loading/None/summary.json"
422
+ }
423
+ },
424
+ "Equi.": {
425
+ "status": "reported",
426
+ "episodes": 50,
427
+ "successes": 10,
428
+ "lscr": 40.666666666666664,
429
+ "source": {
430
+ "runId": "pyrua-dex_r01_21_round_fallback-33066e1d92",
431
+ "perEpisodeFile": "per_task_eval/21_condiment_box_loading/Equi/per_episode.jsonl",
432
+ "summaryFile": "per_task_eval/21_condiment_box_loading/Equi/summary.json"
433
+ }
434
+ },
435
+ "Inv.": {
436
+ "status": "reported",
437
+ "episodes": 50,
438
+ "successes": 9,
439
+ "lscr": 46.0,
440
+ "source": {
441
+ "runId": "pyrua-dex_r01_21_round_fallback-33066e1d92",
442
+ "perEpisodeFile": "per_task_eval/21_condiment_box_loading/Inv/per_episode.jsonl",
443
+ "summaryFile": "per_task_eval/21_condiment_box_loading/Inv/summary.json"
444
+ }
445
+ },
446
+ "Full": {
447
+ "status": "reported",
448
+ "episodes": 50,
449
+ "successes": 14,
450
+ "lscr": 50.0,
451
+ "source": {
452
+ "runId": "pyrua-dex_r01_21_round_fallback-33066e1d92",
453
+ "perEpisodeFile": "per_task_eval/21_condiment_box_loading/Full/per_episode.jsonl",
454
+ "summaryFile": "per_task_eval/21_condiment_box_loading/Full/summary.json"
455
+ }
456
+ }
457
+ },
458
+ "mean": {
459
+ "status": "reported",
460
+ "sr": 20.5,
461
+ "srScope": "four-channel-mean",
462
+ "srEpisodes": 200,
463
+ "lscr": 45.5,
464
+ "lscrScope": "four-channel-mean",
465
+ "lscrEpisodes": 200,
466
+ "efficiency": 0.7647390500248095,
467
+ "efficiencyStatus": "reported"
468
+ }
469
+ },
470
+ "22": {
471
+ "channels": {
472
+ "None": {
473
+ "status": "reported",
474
+ "episodes": 50,
475
+ "successes": 20,
476
+ "lscr": 70.5,
477
+ "source": {
478
+ "runId": "pyrua-dex_r01_22_toolbox-59184eb3a0",
479
+ "perEpisodeFile": "per_task_eval/22_tool_box_loading/None/per_episode.jsonl",
480
+ "summaryFile": "per_task_eval/22_tool_box_loading/None/summary.json"
481
+ }
482
+ },
483
+ "Equi.": {
484
+ "status": "reported",
485
+ "episodes": 50,
486
+ "successes": 26,
487
+ "lscr": 82.0,
488
+ "source": {
489
+ "runId": "pyrua-dex_r01_22_toolbox-59184eb3a0",
490
+ "perEpisodeFile": "per_task_eval/22_tool_box_loading/Equi/per_episode.jsonl",
491
+ "summaryFile": "per_task_eval/22_tool_box_loading/Equi/summary.json"
492
+ }
493
+ },
494
+ "Inv.": {
495
+ "status": "reported",
496
+ "episodes": 50,
497
+ "successes": 25,
498
+ "lscr": 78.0,
499
+ "source": {
500
+ "runId": "pyrua-dex_r01_22_toolbox-59184eb3a0",
501
+ "perEpisodeFile": "per_task_eval/22_tool_box_loading/Inv/per_episode.jsonl",
502
+ "summaryFile": "per_task_eval/22_tool_box_loading/Inv/summary.json"
503
+ }
504
+ },
505
+ "Full": {
506
+ "status": "reported",
507
+ "episodes": 50,
508
+ "successes": 23,
509
+ "lscr": 76.5,
510
+ "source": {
511
+ "runId": "pyrua-dex_r01_22_toolbox-59184eb3a0",
512
+ "perEpisodeFile": "per_task_eval/22_tool_box_loading/Full/per_episode.jsonl",
513
+ "summaryFile": "per_task_eval/22_tool_box_loading/Full/summary.json"
514
+ }
515
+ }
516
+ },
517
+ "mean": {
518
+ "status": "reported",
519
+ "sr": 47.0,
520
+ "srScope": "four-channel-mean",
521
+ "srEpisodes": 200,
522
+ "lscr": 76.75,
523
+ "lscrScope": "four-channel-mean",
524
+ "lscrEpisodes": 200,
525
+ "efficiency": 0.7255332301183569,
526
+ "efficiencyStatus": "reported"
527
+ }
528
+ },
529
+ "24": {
530
+ "channels": {
531
+ "None": {
532
+ "status": "reported",
533
+ "episodes": 50,
534
+ "successes": 2,
535
+ "lscr": 40.0,
536
+ "source": {
537
+ "runId": "pyrua-dex_r01_24_stationery_full-fc6bd676a0",
538
+ "perEpisodeFile": "per_task_eval/24_stationery_category_sorting/None/per_episode.jsonl",
539
+ "summaryFile": "per_task_eval/24_stationery_category_sorting/None/summary.json"
540
+ }
541
+ },
542
+ "Equi.": {
543
+ "status": "reported",
544
+ "episodes": 50,
545
+ "successes": 4,
546
+ "lscr": 45.0,
547
+ "source": {
548
+ "runId": "pyrua-dex_r01_24_stationery_full-fc6bd676a0",
549
+ "perEpisodeFile": "per_task_eval/24_stationery_category_sorting/Equi/per_episode.jsonl",
550
+ "summaryFile": "per_task_eval/24_stationery_category_sorting/Equi/summary.json"
551
+ }
552
+ },
553
+ "Inv.": {
554
+ "status": "reported",
555
+ "episodes": 50,
556
+ "successes": 4,
557
+ "lscr": 43.0,
558
+ "source": {
559
+ "runId": "pyrua-dex_r01_24_stationery_full-fc6bd676a0",
560
+ "perEpisodeFile": "per_task_eval/24_stationery_category_sorting/Inv/per_episode.jsonl",
561
+ "summaryFile": "per_task_eval/24_stationery_category_sorting/Inv/summary.json"
562
+ }
563
+ },
564
+ "Full": {
565
+ "status": "reported",
566
+ "episodes": 50,
567
+ "successes": 3,
568
+ "lscr": 42.5,
569
+ "source": {
570
+ "runId": "pyrua-dex_r01_24_stationery_full-fc6bd676a0",
571
+ "perEpisodeFile": "per_task_eval/24_stationery_category_sorting/Full/per_episode.jsonl",
572
+ "summaryFile": "per_task_eval/24_stationery_category_sorting/Full/summary.json"
573
+ }
574
+ }
575
+ },
576
+ "mean": {
577
+ "status": "reported",
578
+ "sr": 6.5,
579
+ "srScope": "four-channel-mean",
580
+ "srEpisodes": 200,
581
+ "lscr": 42.625,
582
+ "lscrScope": "four-channel-mean",
583
+ "lscrEpisodes": 200,
584
+ "efficiency": 0.903758991589059,
585
+ "efficiencyStatus": "reported"
586
+ }
587
+ },
588
+ "26": {
589
+ "channels": {
590
+ "None": {
591
+ "status": "reported",
592
+ "episodes": 50,
593
+ "successes": 21,
594
+ "lscr": 83.5,
595
+ "source": {
596
+ "runId": "pyrua-dex_r01_26_tray_round-c2ad9643b5",
597
+ "perEpisodeFile": "per_task_eval/26_canned_food_tray_line_arrangement/None/per_episode.jsonl",
598
+ "summaryFile": "per_task_eval/26_canned_food_tray_line_arrangement/None/summary.json"
599
+ }
600
+ },
601
+ "Equi.": {
602
+ "status": "reported",
603
+ "episodes": 50,
604
+ "successes": 20,
605
+ "lscr": 79.0,
606
+ "source": {
607
+ "runId": "pyrua-dex_r01_26_tray_round-c2ad9643b5",
608
+ "perEpisodeFile": "per_task_eval/26_canned_food_tray_line_arrangement/Equi/per_episode.jsonl",
609
+ "summaryFile": "per_task_eval/26_canned_food_tray_line_arrangement/Equi/summary.json"
610
+ }
611
+ },
612
+ "Inv.": {
613
+ "status": "reported",
614
+ "episodes": 50,
615
+ "successes": 20,
616
+ "lscr": 78.5,
617
+ "source": {
618
+ "runId": "pyrua-dex_r01_26_tray_round-c2ad9643b5",
619
+ "perEpisodeFile": "per_task_eval/26_canned_food_tray_line_arrangement/Inv/per_episode.jsonl",
620
+ "summaryFile": "per_task_eval/26_canned_food_tray_line_arrangement/Inv/summary.json"
621
+ }
622
+ },
623
+ "Full": {
624
+ "status": "reported",
625
+ "episodes": 50,
626
+ "successes": 22,
627
+ "lscr": 80.0,
628
+ "source": {
629
+ "runId": "pyrua-dex_r01_26_tray_round-c2ad9643b5",
630
+ "perEpisodeFile": "per_task_eval/26_canned_food_tray_line_arrangement/Full/per_episode.jsonl",
631
+ "summaryFile": "per_task_eval/26_canned_food_tray_line_arrangement/Full/summary.json"
632
+ }
633
+ }
634
+ },
635
+ "mean": {
636
+ "status": "reported",
637
+ "sr": 41.5,
638
+ "srScope": "four-channel-mean",
639
+ "srEpisodes": 200,
640
+ "lscr": 80.25,
641
+ "lscrScope": "four-channel-mean",
642
+ "lscrEpisodes": 200,
643
+ "efficiency": 1.0564904460925797,
644
+ "efficiencyStatus": "reported"
645
+ }
646
+ },
647
+ "27": {
648
+ "channels": {
649
+ "None": {
650
+ "status": "reported",
651
+ "episodes": 50,
652
+ "successes": 41,
653
+ "lscr": 93.5,
654
+ "source": {
655
+ "runId": "pyrua-dex_r01_27_ring_prefix_thumb_release-f1ac959cf9",
656
+ "perEpisodeFile": "per_task_eval/27_ball_box_loading/None/per_episode.jsonl",
657
+ "summaryFile": "per_task_eval/27_ball_box_loading/None/summary.json"
658
+ }
659
+ },
660
+ "Equi.": {
661
+ "status": "reported",
662
+ "episodes": 50,
663
+ "successes": 39,
664
+ "lscr": 92.5,
665
+ "source": {
666
+ "runId": "pyrua-dex_r01_27_ring_prefix_thumb_release-f1ac959cf9",
667
+ "perEpisodeFile": "per_task_eval/27_ball_box_loading/Equi/per_episode.jsonl",
668
+ "summaryFile": "per_task_eval/27_ball_box_loading/Equi/summary.json"
669
+ }
670
+ },
671
+ "Inv.": {
672
+ "status": "reported",
673
+ "episodes": 50,
674
+ "successes": 34,
675
+ "lscr": 90.5,
676
+ "source": {
677
+ "runId": "pyrua-dex_r01_27_ring_prefix_thumb_release-f1ac959cf9",
678
+ "perEpisodeFile": "per_task_eval/27_ball_box_loading/Inv/per_episode.jsonl",
679
+ "summaryFile": "per_task_eval/27_ball_box_loading/Inv/summary.json"
680
+ }
681
+ },
682
+ "Full": {
683
+ "status": "reported",
684
+ "episodes": 50,
685
+ "successes": 35,
686
+ "lscr": 89.0,
687
+ "source": {
688
+ "runId": "pyrua-dex_r01_27_ring_prefix_thumb_release-f1ac959cf9",
689
+ "perEpisodeFile": "per_task_eval/27_ball_box_loading/Full/per_episode.jsonl",
690
+ "summaryFile": "per_task_eval/27_ball_box_loading/Full/summary.json"
691
+ }
692
+ }
693
+ },
694
+ "mean": {
695
+ "status": "reported",
696
+ "sr": 74.5,
697
+ "srScope": "four-channel-mean",
698
+ "srEpisodes": 200,
699
+ "lscr": 91.375,
700
+ "lscrScope": "four-channel-mean",
701
+ "lscrEpisodes": 200,
702
+ "efficiency": 0.7555346288480345,
703
+ "efficiencyStatus": "reported"
704
+ }
705
+ },
706
+ "32": {
707
+ "channels": {
708
+ "None": {
709
+ "status": "reported",
710
+ "episodes": 50,
711
+ "successes": 13,
712
+ "lscr": 72.5,
713
+ "source": {
714
+ "runId": "pyrua-dex_r01_32_tray_hybrid-c636840313",
715
+ "perEpisodeFile": "per_task_eval/32_baking_tray_prep_with_tools/None/per_episode.jsonl",
716
+ "summaryFile": "per_task_eval/32_baking_tray_prep_with_tools/None/summary.json"
717
+ }
718
+ },
719
+ "Equi.": {
720
+ "status": "reported",
721
+ "episodes": 50,
722
+ "successes": 9,
723
+ "lscr": 62.5,
724
+ "source": {
725
+ "runId": "pyrua-dex_r01_32_tray_hybrid-c636840313",
726
+ "perEpisodeFile": "per_task_eval/32_baking_tray_prep_with_tools/Equi/per_episode.jsonl",
727
+ "summaryFile": "per_task_eval/32_baking_tray_prep_with_tools/Equi/summary.json"
728
+ }
729
+ },
730
+ "Inv.": {
731
+ "status": "reported",
732
+ "episodes": 50,
733
+ "successes": 19,
734
+ "lscr": 78.0,
735
+ "source": {
736
+ "runId": "pyrua-dex_r01_32_tray_hybrid-c636840313",
737
+ "perEpisodeFile": "per_task_eval/32_baking_tray_prep_with_tools/Inv/per_episode.jsonl",
738
+ "summaryFile": "per_task_eval/32_baking_tray_prep_with_tools/Inv/summary.json"
739
+ }
740
+ },
741
+ "Full": {
742
+ "status": "reported",
743
+ "episodes": 50,
744
+ "successes": 11,
745
+ "lscr": 70.5,
746
+ "source": {
747
+ "runId": "pyrua-dex_r01_32_tray_hybrid-c636840313",
748
+ "perEpisodeFile": "per_task_eval/32_baking_tray_prep_with_tools/Full/per_episode.jsonl",
749
+ "summaryFile": "per_task_eval/32_baking_tray_prep_with_tools/Full/summary.json"
750
+ }
751
+ }
752
+ },
753
+ "mean": {
754
+ "status": "reported",
755
+ "sr": 26.0,
756
+ "srScope": "four-channel-mean",
757
+ "srEpisodes": 200,
758
+ "lscr": 70.875,
759
+ "lscrScope": "four-channel-mean",
760
+ "lscrEpisodes": 200,
761
+ "efficiency": 0.740657723060453,
762
+ "efficiencyStatus": "reported"
763
+ }
764
+ },
765
+ "34": {
766
+ "channels": {
767
+ "None": {
768
+ "status": "reported",
769
+ "episodes": 50,
770
+ "successes": 0,
771
+ "lscr": 40.800000000000004,
772
+ "source": {
773
+ "runId": "pyrua-dex_r01_34_retensioned_service-3cc32133ca",
774
+ "perEpisodeFile": "per_task_eval/34_fridge_wine_interhand_pour/None/per_episode.jsonl",
775
+ "summaryFile": "per_task_eval/34_fridge_wine_interhand_pour/None/summary.json"
776
+ }
777
+ },
778
+ "Equi.": {
779
+ "status": "reported",
780
+ "episodes": 50,
781
+ "successes": 0,
782
+ "lscr": 40.800000000000004,
783
+ "source": {
784
+ "runId": "pyrua-dex_r01_34_retensioned_service-3cc32133ca",
785
+ "perEpisodeFile": "per_task_eval/34_fridge_wine_interhand_pour/Equi/per_episode.jsonl",
786
+ "summaryFile": "per_task_eval/34_fridge_wine_interhand_pour/Equi/summary.json"
787
+ }
788
+ },
789
+ "Inv.": {
790
+ "status": "reported",
791
+ "episodes": 50,
792
+ "successes": 0,
793
+ "lscr": 40.0,
794
+ "source": {
795
+ "runId": "pyrua-dex_r01_34_retensioned_service-3cc32133ca",
796
+ "perEpisodeFile": "per_task_eval/34_fridge_wine_interhand_pour/Inv/per_episode.jsonl",
797
+ "summaryFile": "per_task_eval/34_fridge_wine_interhand_pour/Inv/summary.json"
798
+ }
799
+ },
800
+ "Full": {
801
+ "status": "reported",
802
+ "episodes": 50,
803
+ "successes": 0,
804
+ "lscr": 39.6,
805
+ "source": {
806
+ "runId": "pyrua-dex_r01_34_retensioned_service-3cc32133ca",
807
+ "perEpisodeFile": "per_task_eval/34_fridge_wine_interhand_pour/Full/per_episode.jsonl",
808
+ "summaryFile": "per_task_eval/34_fridge_wine_interhand_pour/Full/summary.json"
809
+ }
810
+ }
811
+ },
812
+ "mean": {
813
+ "status": "reported",
814
+ "sr": 0.0,
815
+ "srScope": "four-channel-mean",
816
+ "srEpisodes": 200,
817
+ "lscr": 40.300000000000004,
818
+ "lscrScope": "four-channel-mean",
819
+ "lscrEpisodes": 200,
820
+ "efficiency": null,
821
+ "efficiencyStatus": "no-success"
822
+ }
823
+ },
824
+ "42": {
825
+ "channels": {
826
+ "None": {
827
+ "status": "reported",
828
+ "episodes": 50,
829
+ "successes": 22,
830
+ "lscr": 66.66666666666666,
831
+ "source": {
832
+ "runId": "pyrua-dex_r01_42_trash-f39e0082b0",
833
+ "perEpisodeFile": "per_task_eval/42_trash_disposal/None/per_episode.jsonl",
834
+ "summaryFile": "per_task_eval/42_trash_disposal/None/summary.json"
835
+ }
836
+ },
837
+ "Equi.": {
838
+ "status": "reported",
839
+ "episodes": 50,
840
+ "successes": 20,
841
+ "lscr": 69.33333333333334,
842
+ "source": {
843
+ "runId": "pyrua-dex_r01_42_trash-f39e0082b0",
844
+ "perEpisodeFile": "per_task_eval/42_trash_disposal/Equi/per_episode.jsonl",
845
+ "summaryFile": "per_task_eval/42_trash_disposal/Equi/summary.json"
846
+ }
847
+ },
848
+ "Inv.": {
849
+ "status": "reported",
850
+ "episodes": 50,
851
+ "successes": 20,
852
+ "lscr": 63.33333333333333,
853
+ "source": {
854
+ "runId": "pyrua-dex_r01_42_trash-f39e0082b0",
855
+ "perEpisodeFile": "per_task_eval/42_trash_disposal/Inv/per_episode.jsonl",
856
+ "summaryFile": "per_task_eval/42_trash_disposal/Inv/summary.json"
857
+ }
858
+ },
859
+ "Full": {
860
+ "status": "reported",
861
+ "episodes": 50,
862
+ "successes": 17,
863
+ "lscr": 67.33333333333333,
864
+ "source": {
865
+ "runId": "pyrua-dex_r01_42_trash-f39e0082b0",
866
+ "perEpisodeFile": "per_task_eval/42_trash_disposal/Full/per_episode.jsonl",
867
+ "summaryFile": "per_task_eval/42_trash_disposal/Full/summary.json"
868
+ }
869
+ }
870
+ },
871
+ "mean": {
872
+ "status": "reported",
873
+ "sr": 39.5,
874
+ "srScope": "four-channel-mean",
875
+ "srEpisodes": 200,
876
+ "lscr": 66.66666666666666,
877
+ "lscrScope": "four-channel-mean",
878
+ "lscrEpisodes": 200,
879
+ "efficiency": 0.7852041423640043,
880
+ "efficiencyStatus": "reported"
881
+ }
882
+ },
883
+ "43": {
884
+ "channels": {
885
+ "None": {
886
+ "status": "reported",
887
+ "episodes": 50,
888
+ "successes": 2,
889
+ "lscr": 44.5,
890
+ "source": {
891
+ "runId": "pyrua-dex_r01_43_fridge_round-c4e0b828aa",
892
+ "perEpisodeFile": "per_task_eval/43_fridge_fruit_shelf_sorting/None/per_episode.jsonl",
893
+ "summaryFile": "per_task_eval/43_fridge_fruit_shelf_sorting/None/summary.json"
894
+ }
895
+ },
896
+ "Equi.": {
897
+ "status": "reported",
898
+ "episodes": 50,
899
+ "successes": 7,
900
+ "lscr": 49.5,
901
+ "source": {
902
+ "runId": "pyrua-dex_r01_43_fridge_round-c4e0b828aa",
903
+ "perEpisodeFile": "per_task_eval/43_fridge_fruit_shelf_sorting/Equi/per_episode.jsonl",
904
+ "summaryFile": "per_task_eval/43_fridge_fruit_shelf_sorting/Equi/summary.json"
905
+ }
906
+ },
907
+ "Inv.": {
908
+ "status": "reported",
909
+ "episodes": 50,
910
+ "successes": 2,
911
+ "lscr": 38.5,
912
+ "source": {
913
+ "runId": "pyrua-dex_r01_43_fridge_round-c4e0b828aa",
914
+ "perEpisodeFile": "per_task_eval/43_fridge_fruit_shelf_sorting/Inv/per_episode.jsonl",
915
+ "summaryFile": "per_task_eval/43_fridge_fruit_shelf_sorting/Inv/summary.json"
916
+ }
917
+ },
918
+ "Full": {
919
+ "status": "reported",
920
+ "episodes": 50,
921
+ "successes": 4,
922
+ "lscr": 45.0,
923
+ "source": {
924
+ "runId": "pyrua-dex_r01_43_fridge_round-c4e0b828aa",
925
+ "perEpisodeFile": "per_task_eval/43_fridge_fruit_shelf_sorting/Full/per_episode.jsonl",
926
+ "summaryFile": "per_task_eval/43_fridge_fruit_shelf_sorting/Full/summary.json"
927
+ }
928
+ }
929
+ },
930
+ "mean": {
931
+ "status": "reported",
932
+ "sr": 7.5,
933
+ "srScope": "four-channel-mean",
934
+ "srEpisodes": 200,
935
+ "lscr": 44.375,
936
+ "lscrScope": "four-channel-mean",
937
+ "lscrEpisodes": 200,
938
+ "efficiency": 0.7212042800195818,
939
+ "efficiencyStatus": "reported"
940
+ }
941
+ },
942
+ "44": {
943
+ "channels": {
944
+ "None": {
945
+ "status": "reported",
946
+ "episodes": 50,
947
+ "successes": 32,
948
+ "lscr": 74.5,
949
+ "source": {
950
+ "runId": "pyrua-dex_r01_44_reliable-a666046e78",
951
+ "perEpisodeFile": "per_task_eval/44_microwave_bowl_loading/None/per_episode.jsonl",
952
+ "summaryFile": "per_task_eval/44_microwave_bowl_loading/None/summary.json"
953
+ }
954
+ },
955
+ "Equi.": {
956
+ "status": "reported",
957
+ "episodes": 50,
958
+ "successes": 33,
959
+ "lscr": 77.0,
960
+ "source": {
961
+ "runId": "pyrua-dex_r01_44_reliable-a666046e78",
962
+ "perEpisodeFile": "per_task_eval/44_microwave_bowl_loading/Equi/per_episode.jsonl",
963
+ "summaryFile": "per_task_eval/44_microwave_bowl_loading/Equi/summary.json"
964
+ }
965
+ },
966
+ "Inv.": {
967
+ "status": "reported",
968
+ "episodes": 50,
969
+ "successes": 31,
970
+ "lscr": 74.5,
971
+ "source": {
972
+ "runId": "pyrua-dex_r01_44_reliable-a666046e78",
973
+ "perEpisodeFile": "per_task_eval/44_microwave_bowl_loading/Inv/per_episode.jsonl",
974
+ "summaryFile": "per_task_eval/44_microwave_bowl_loading/Inv/summary.json"
975
+ }
976
+ },
977
+ "Full": {
978
+ "status": "reported",
979
+ "episodes": 50,
980
+ "successes": 34,
981
+ "lscr": 78.5,
982
+ "source": {
983
+ "runId": "pyrua-dex_r01_44_reliable-a666046e78",
984
+ "perEpisodeFile": "per_task_eval/44_microwave_bowl_loading/Full/per_episode.jsonl",
985
+ "summaryFile": "per_task_eval/44_microwave_bowl_loading/Full/summary.json"
986
+ }
987
+ }
988
+ },
989
+ "mean": {
990
+ "status": "reported",
991
+ "sr": 65.0,
992
+ "srScope": "four-channel-mean",
993
+ "srEpisodes": 200,
994
+ "lscr": 76.125,
995
+ "lscrScope": "four-channel-mean",
996
+ "lscrEpisodes": 200,
997
+ "efficiency": 0.8862630195783616,
998
+ "efficiencyStatus": "reported"
999
+ }
1000
+ },
1001
+ "51": {
1002
+ "channels": {
1003
+ "None": {
1004
+ "status": "reported",
1005
+ "episodes": 50,
1006
+ "successes": 42,
1007
+ "lscr": 89.33333333333333,
1008
+ "source": {
1009
+ "runId": "pyrua-dex_r01_51_toilet_final-4a83db3920",
1010
+ "perEpisodeFile": "per_task_eval/51_toilet_lid_cleaner_pour/None/per_episode.jsonl",
1011
+ "summaryFile": "per_task_eval/51_toilet_lid_cleaner_pour/None/summary.json"
1012
+ }
1013
+ },
1014
+ "Equi.": {
1015
+ "status": "reported",
1016
+ "episodes": 50,
1017
+ "successes": 45,
1018
+ "lscr": 92.66666666666666,
1019
+ "source": {
1020
+ "runId": "pyrua-dex_r01_51_toilet_final-4a83db3920",
1021
+ "perEpisodeFile": "per_task_eval/51_toilet_lid_cleaner_pour/Equi/per_episode.jsonl",
1022
+ "summaryFile": "per_task_eval/51_toilet_lid_cleaner_pour/Equi/summary.json"
1023
+ }
1024
+ },
1025
+ "Inv.": {
1026
+ "status": "reported",
1027
+ "episodes": 50,
1028
+ "successes": 44,
1029
+ "lscr": 90.66666666666666,
1030
+ "source": {
1031
+ "runId": "pyrua-dex_r01_51_toilet_final-4a83db3920",
1032
+ "perEpisodeFile": "per_task_eval/51_toilet_lid_cleaner_pour/Inv/per_episode.jsonl",
1033
+ "summaryFile": "per_task_eval/51_toilet_lid_cleaner_pour/Inv/summary.json"
1034
+ }
1035
+ },
1036
+ "Full": {
1037
+ "status": "reported",
1038
+ "episodes": 50,
1039
+ "successes": 40,
1040
+ "lscr": 87.33333333333333,
1041
+ "source": {
1042
+ "runId": "pyrua-dex_r01_51_toilet_final-4a83db3920",
1043
+ "perEpisodeFile": "per_task_eval/51_toilet_lid_cleaner_pour/Full/per_episode.jsonl",
1044
+ "summaryFile": "per_task_eval/51_toilet_lid_cleaner_pour/Full/summary.json"
1045
+ }
1046
+ }
1047
+ },
1048
+ "mean": {
1049
+ "status": "reported",
1050
+ "sr": 85.5,
1051
+ "srScope": "four-channel-mean",
1052
+ "srEpisodes": 200,
1053
+ "lscr": 90.0,
1054
+ "lscrScope": "four-channel-mean",
1055
+ "lscrEpisodes": 200,
1056
+ "efficiency": 1.4939918524907978,
1057
+ "efficiencyStatus": "reported"
1058
+ }
1059
+ },
1060
+ "60": {
1061
+ "channels": {
1062
+ "None": {
1063
+ "status": "reported",
1064
+ "episodes": 50,
1065
+ "successes": 26,
1066
+ "lscr": 61.0,
1067
+ "source": {
1068
+ "runId": "pyrua-dex_r01_60_meal-560f17c29a",
1069
+ "perEpisodeFile": "per_task_eval/60_breadbasket_fast_food_loading/None/per_episode.jsonl",
1070
+ "summaryFile": "per_task_eval/60_breadbasket_fast_food_loading/None/summary.json"
1071
+ }
1072
+ },
1073
+ "Equi.": {
1074
+ "status": "reported",
1075
+ "episodes": 50,
1076
+ "successes": 22,
1077
+ "lscr": 50.0,
1078
+ "source": {
1079
+ "runId": "pyrua-dex_r01_60_meal-560f17c29a",
1080
+ "perEpisodeFile": "per_task_eval/60_breadbasket_fast_food_loading/Equi/per_episode.jsonl",
1081
+ "summaryFile": "per_task_eval/60_breadbasket_fast_food_loading/Equi/summary.json"
1082
+ }
1083
+ },
1084
+ "Inv.": {
1085
+ "status": "reported",
1086
+ "episodes": 50,
1087
+ "successes": 24,
1088
+ "lscr": 52.5,
1089
+ "source": {
1090
+ "runId": "pyrua-dex_r01_60_meal-560f17c29a",
1091
+ "perEpisodeFile": "per_task_eval/60_breadbasket_fast_food_loading/Inv/per_episode.jsonl",
1092
+ "summaryFile": "per_task_eval/60_breadbasket_fast_food_loading/Inv/summary.json"
1093
+ }
1094
+ },
1095
+ "Full": {
1096
+ "status": "reported",
1097
+ "episodes": 50,
1098
+ "successes": 20,
1099
+ "lscr": 52.0,
1100
+ "source": {
1101
+ "runId": "pyrua-dex_r01_60_meal-560f17c29a",
1102
+ "perEpisodeFile": "per_task_eval/60_breadbasket_fast_food_loading/Full/per_episode.jsonl",
1103
+ "summaryFile": "per_task_eval/60_breadbasket_fast_food_loading/Full/summary.json"
1104
+ }
1105
+ }
1106
+ },
1107
+ "mean": {
1108
+ "status": "reported",
1109
+ "sr": 46.0,
1110
+ "srScope": "four-channel-mean",
1111
+ "srEpisodes": 200,
1112
+ "lscr": 53.875,
1113
+ "lscrScope": "four-channel-mean",
1114
+ "lscrEpisodes": 200,
1115
+ "efficiency": 0.8785377868037851,
1116
+ "efficiencyStatus": "reported"
1117
+ }
1118
+ },
1119
+ "61": {
1120
+ "channels": {
1121
+ "None": {
1122
+ "status": "reported",
1123
+ "episodes": 50,
1124
+ "successes": 41,
1125
+ "lscr": 88.0,
1126
+ "source": {
1127
+ "runId": "pyrua-dex_r01_61_medicine_point_v2-06e06faffc",
1128
+ "perEpisodeFile": "per_task_eval/61_medicine_shoebox_pack/None/per_episode.jsonl",
1129
+ "summaryFile": "per_task_eval/61_medicine_shoebox_pack/None/summary.json"
1130
+ }
1131
+ },
1132
+ "Equi.": {
1133
+ "status": "reported",
1134
+ "episodes": 50,
1135
+ "successes": 38,
1136
+ "lscr": 86.0,
1137
+ "source": {
1138
+ "runId": "pyrua-dex_r01_61_medicine_point_v2-06e06faffc",
1139
+ "perEpisodeFile": "per_task_eval/61_medicine_shoebox_pack/Equi/per_episode.jsonl",
1140
+ "summaryFile": "per_task_eval/61_medicine_shoebox_pack/Equi/summary.json"
1141
+ }
1142
+ },
1143
+ "Inv.": {
1144
+ "status": "reported",
1145
+ "episodes": 50,
1146
+ "successes": 42,
1147
+ "lscr": 90.0,
1148
+ "source": {
1149
+ "runId": "pyrua-dex_r01_61_medicine_point_v2-06e06faffc",
1150
+ "perEpisodeFile": "per_task_eval/61_medicine_shoebox_pack/Inv/per_episode.jsonl",
1151
+ "summaryFile": "per_task_eval/61_medicine_shoebox_pack/Inv/summary.json"
1152
+ }
1153
+ },
1154
+ "Full": {
1155
+ "status": "reported",
1156
+ "episodes": 50,
1157
+ "successes": 38,
1158
+ "lscr": 87.33333333333333,
1159
+ "source": {
1160
+ "runId": "pyrua-dex_r01_61_medicine_point_v2-06e06faffc",
1161
+ "perEpisodeFile": "per_task_eval/61_medicine_shoebox_pack/Full/per_episode.jsonl",
1162
+ "summaryFile": "per_task_eval/61_medicine_shoebox_pack/Full/summary.json"
1163
+ }
1164
+ }
1165
+ },
1166
+ "mean": {
1167
+ "status": "reported",
1168
+ "sr": 79.5,
1169
+ "srScope": "four-channel-mean",
1170
+ "srEpisodes": 200,
1171
+ "lscr": 87.83333333333333,
1172
+ "lscrScope": "four-channel-mean",
1173
+ "lscrEpisodes": 200,
1174
+ "efficiency": 0.9014564704591584,
1175
+ "efficiencyStatus": "reported"
1176
+ }
1177
+ },
1178
+ "62": {
1179
+ "channels": {
1180
+ "None": {
1181
+ "status": "reported",
1182
+ "episodes": 50,
1183
+ "successes": 26,
1184
+ "lscr": 65.33333333333333,
1185
+ "source": {
1186
+ "runId": "pyrua-dex_r01_62_shoebox-8b45ce4db8",
1187
+ "perEpisodeFile": "per_task_eval/62_shoebox_accessory_pack/None/per_episode.jsonl",
1188
+ "summaryFile": "per_task_eval/62_shoebox_accessory_pack/None/summary.json"
1189
+ }
1190
+ },
1191
+ "Equi.": {
1192
+ "status": "reported",
1193
+ "episodes": 50,
1194
+ "successes": 32,
1195
+ "lscr": 76.66666666666666,
1196
+ "source": {
1197
+ "runId": "pyrua-dex_r01_62_shoebox-8b45ce4db8",
1198
+ "perEpisodeFile": "per_task_eval/62_shoebox_accessory_pack/Equi/per_episode.jsonl",
1199
+ "summaryFile": "per_task_eval/62_shoebox_accessory_pack/Equi/summary.json"
1200
+ }
1201
+ },
1202
+ "Inv.": {
1203
+ "status": "reported",
1204
+ "episodes": 50,
1205
+ "successes": 23,
1206
+ "lscr": 64.66666666666666,
1207
+ "source": {
1208
+ "runId": "pyrua-dex_r01_62_shoebox-8b45ce4db8",
1209
+ "perEpisodeFile": "per_task_eval/62_shoebox_accessory_pack/Inv/per_episode.jsonl",
1210
+ "summaryFile": "per_task_eval/62_shoebox_accessory_pack/Inv/summary.json"
1211
+ }
1212
+ },
1213
+ "Full": {
1214
+ "status": "reported",
1215
+ "episodes": 50,
1216
+ "successes": 29,
1217
+ "lscr": 66.0,
1218
+ "source": {
1219
+ "runId": "pyrua-dex_r01_62_shoebox-8b45ce4db8",
1220
+ "perEpisodeFile": "per_task_eval/62_shoebox_accessory_pack/Full/per_episode.jsonl",
1221
+ "summaryFile": "per_task_eval/62_shoebox_accessory_pack/Full/summary.json"
1222
+ }
1223
+ }
1224
+ },
1225
+ "mean": {
1226
+ "status": "reported",
1227
+ "sr": 55.0,
1228
+ "srScope": "four-channel-mean",
1229
+ "srEpisodes": 200,
1230
+ "lscr": 68.16666666666666,
1231
+ "lscrScope": "four-channel-mean",
1232
+ "lscrEpisodes": 200,
1233
+ "efficiency": 0.770143373952161,
1234
+ "efficiencyStatus": "reported"
1235
+ }
1236
+ },
1237
+ "64": {
1238
+ "channels": {
1239
+ "None": {
1240
+ "status": "reported",
1241
+ "episodes": 50,
1242
+ "successes": 21,
1243
+ "lscr": 53.5,
1244
+ "source": {
1245
+ "runId": "pyrua-dex_r01_64_ball_sort-7a836c8a10",
1246
+ "perEpisodeFile": "per_task_eval/64_sports_ball_cup_sort/None/per_episode.jsonl",
1247
+ "summaryFile": "per_task_eval/64_sports_ball_cup_sort/None/summary.json"
1248
+ }
1249
+ },
1250
+ "Equi.": {
1251
+ "status": "reported",
1252
+ "episodes": 50,
1253
+ "successes": 26,
1254
+ "lscr": 64.0,
1255
+ "source": {
1256
+ "runId": "pyrua-dex_r01_64_ball_sort-7a836c8a10",
1257
+ "perEpisodeFile": "per_task_eval/64_sports_ball_cup_sort/Equi/per_episode.jsonl",
1258
+ "summaryFile": "per_task_eval/64_sports_ball_cup_sort/Equi/summary.json"
1259
+ }
1260
+ },
1261
+ "Inv.": {
1262
+ "status": "reported",
1263
+ "episodes": 50,
1264
+ "successes": 28,
1265
+ "lscr": 64.5,
1266
+ "source": {
1267
+ "runId": "pyrua-dex_r01_64_ball_sort-7a836c8a10",
1268
+ "perEpisodeFile": "per_task_eval/64_sports_ball_cup_sort/Inv/per_episode.jsonl",
1269
+ "summaryFile": "per_task_eval/64_sports_ball_cup_sort/Inv/summary.json"
1270
+ }
1271
+ },
1272
+ "Full": {
1273
+ "status": "reported",
1274
+ "episodes": 50,
1275
+ "successes": 19,
1276
+ "lscr": 53.0,
1277
+ "source": {
1278
+ "runId": "pyrua-dex_r01_64_ball_sort-7a836c8a10",
1279
+ "perEpisodeFile": "per_task_eval/64_sports_ball_cup_sort/Full/per_episode.jsonl",
1280
+ "summaryFile": "per_task_eval/64_sports_ball_cup_sort/Full/summary.json"
1281
+ }
1282
+ }
1283
+ },
1284
+ "mean": {
1285
+ "status": "reported",
1286
+ "sr": 47.0,
1287
+ "srScope": "four-channel-mean",
1288
+ "srEpisodes": 200,
1289
+ "lscr": 58.75,
1290
+ "lscrScope": "four-channel-mean",
1291
+ "lscrEpisodes": 200,
1292
+ "efficiency": 0.792743056216566,
1293
+ "efficiencyStatus": "reported"
1294
+ }
1295
+ },
1296
+ "67": {
1297
+ "channels": {
1298
+ "None": {
1299
+ "status": "reported",
1300
+ "episodes": 50,
1301
+ "successes": 28,
1302
+ "lscr": 72.0,
1303
+ "source": {
1304
+ "runId": "pyrua-dex_r01_67_faucet_recovery-ee8f8b886b",
1305
+ "perEpisodeFile": "per_task_eval/67_faucet_cup_water_fill/None/per_episode.jsonl",
1306
+ "summaryFile": "per_task_eval/67_faucet_cup_water_fill/None/summary.json"
1307
+ }
1308
+ },
1309
+ "Equi.": {
1310
+ "status": "reported",
1311
+ "episodes": 50,
1312
+ "successes": 36,
1313
+ "lscr": 84.0,
1314
+ "source": {
1315
+ "runId": "pyrua-dex_r01_67_faucet_recovery-ee8f8b886b",
1316
+ "perEpisodeFile": "per_task_eval/67_faucet_cup_water_fill/Equi/per_episode.jsonl",
1317
+ "summaryFile": "per_task_eval/67_faucet_cup_water_fill/Equi/summary.json"
1318
+ }
1319
+ },
1320
+ "Inv.": {
1321
+ "status": "reported",
1322
+ "episodes": 50,
1323
+ "successes": 31,
1324
+ "lscr": 74.0,
1325
+ "source": {
1326
+ "runId": "pyrua-dex_r01_67_faucet_recovery-ee8f8b886b",
1327
+ "perEpisodeFile": "per_task_eval/67_faucet_cup_water_fill/Inv/per_episode.jsonl",
1328
+ "summaryFile": "per_task_eval/67_faucet_cup_water_fill/Inv/summary.json"
1329
+ }
1330
+ },
1331
+ "Full": {
1332
+ "status": "reported",
1333
+ "episodes": 50,
1334
+ "successes": 26,
1335
+ "lscr": 67.5,
1336
+ "source": {
1337
+ "runId": "pyrua-dex_r01_67_faucet_recovery-ee8f8b886b",
1338
+ "perEpisodeFile": "per_task_eval/67_faucet_cup_water_fill/Full/per_episode.jsonl",
1339
+ "summaryFile": "per_task_eval/67_faucet_cup_water_fill/Full/summary.json"
1340
+ }
1341
+ }
1342
+ },
1343
+ "mean": {
1344
+ "status": "reported",
1345
+ "sr": 60.5,
1346
+ "srScope": "four-channel-mean",
1347
+ "srEpisodes": 200,
1348
+ "lscr": 74.375,
1349
+ "lscrScope": "four-channel-mean",
1350
+ "lscrEpisodes": 200,
1351
+ "efficiency": 0.7813877463524431,
1352
+ "efficiencyStatus": "reported"
1353
+ }
1354
+ },
1355
+ "73": {
1356
+ "channels": {
1357
+ "None": {
1358
+ "status": "reported",
1359
+ "episodes": 50,
1360
+ "successes": 0,
1361
+ "lscr": 36.0,
1362
+ "source": {
1363
+ "runId": "pyrua-dex_r01_73_yellowbounds-df764f33e4",
1364
+ "perEpisodeFile": "per_task_eval/73_jigsaw_puzzle_assembly/None/per_episode.jsonl",
1365
+ "summaryFile": "per_task_eval/73_jigsaw_puzzle_assembly/None/summary.json"
1366
+ }
1367
+ },
1368
+ "Equi.": {
1369
+ "status": "reported",
1370
+ "episodes": 50,
1371
+ "successes": 0,
1372
+ "lscr": 37.0,
1373
+ "source": {
1374
+ "runId": "pyrua-dex_r01_73_yellowbounds-df764f33e4",
1375
+ "perEpisodeFile": "per_task_eval/73_jigsaw_puzzle_assembly/Equi/per_episode.jsonl",
1376
+ "summaryFile": "per_task_eval/73_jigsaw_puzzle_assembly/Equi/summary.json"
1377
+ }
1378
+ },
1379
+ "Inv.": {
1380
+ "status": "reported",
1381
+ "episodes": 50,
1382
+ "successes": 0,
1383
+ "lscr": 38.0,
1384
+ "source": {
1385
+ "runId": "pyrua-dex_r01_73_yellowbounds-df764f33e4",
1386
+ "perEpisodeFile": "per_task_eval/73_jigsaw_puzzle_assembly/Inv/per_episode.jsonl",
1387
+ "summaryFile": "per_task_eval/73_jigsaw_puzzle_assembly/Inv/summary.json"
1388
+ }
1389
+ },
1390
+ "Full": {
1391
+ "status": "reported",
1392
+ "episodes": 50,
1393
+ "successes": 0,
1394
+ "lscr": 37.0,
1395
+ "source": {
1396
+ "runId": "pyrua-dex_r01_73_yellowbounds-df764f33e4",
1397
+ "perEpisodeFile": "per_task_eval/73_jigsaw_puzzle_assembly/Full/per_episode.jsonl",
1398
+ "summaryFile": "per_task_eval/73_jigsaw_puzzle_assembly/Full/summary.json"
1399
+ }
1400
+ }
1401
+ },
1402
+ "mean": {
1403
+ "status": "reported",
1404
+ "sr": 0.0,
1405
+ "srScope": "four-channel-mean",
1406
+ "srEpisodes": 200,
1407
+ "lscr": 37.0,
1408
+ "lscrScope": "four-channel-mean",
1409
+ "lscrEpisodes": 200,
1410
+ "efficiency": null,
1411
+ "efficiencyStatus": "no-success"
1412
+ }
1413
+ },
1414
+ "76": {
1415
+ "channels": {
1416
+ "None": {
1417
+ "status": "reported",
1418
+ "episodes": 50,
1419
+ "successes": 9,
1420
+ "lscr": 27.0,
1421
+ "source": {
1422
+ "runId": "pyrua-dex_r01_76_serving_palm-139449a835",
1423
+ "perEpisodeFile": "per_task_eval/76_soup_serving/None/per_episode.jsonl",
1424
+ "summaryFile": "per_task_eval/76_soup_serving/None/summary.json"
1425
+ }
1426
+ },
1427
+ "Equi.": {
1428
+ "status": "reported",
1429
+ "episodes": 50,
1430
+ "successes": 6,
1431
+ "lscr": 22.5,
1432
+ "source": {
1433
+ "runId": "pyrua-dex_r01_76_serving_palm-139449a835",
1434
+ "perEpisodeFile": "per_task_eval/76_soup_serving/Equi/per_episode.jsonl",
1435
+ "summaryFile": "per_task_eval/76_soup_serving/Equi/summary.json"
1436
+ }
1437
+ },
1438
+ "Inv.": {
1439
+ "status": "reported",
1440
+ "episodes": 50,
1441
+ "successes": 5,
1442
+ "lscr": 27.500000000000004,
1443
+ "source": {
1444
+ "runId": "pyrua-dex_r01_76_serving_palm-139449a835",
1445
+ "perEpisodeFile": "per_task_eval/76_soup_serving/Inv/per_episode.jsonl",
1446
+ "summaryFile": "per_task_eval/76_soup_serving/Inv/summary.json"
1447
+ }
1448
+ },
1449
+ "Full": {
1450
+ "status": "reported",
1451
+ "episodes": 50,
1452
+ "successes": 8,
1453
+ "lscr": 23.0,
1454
+ "source": {
1455
+ "runId": "pyrua-dex_r01_76_serving_palm-139449a835",
1456
+ "perEpisodeFile": "per_task_eval/76_soup_serving/Full/per_episode.jsonl",
1457
+ "summaryFile": "per_task_eval/76_soup_serving/Full/summary.json"
1458
+ }
1459
+ }
1460
+ },
1461
+ "mean": {
1462
+ "status": "reported",
1463
+ "sr": 14.0,
1464
+ "srScope": "four-channel-mean",
1465
+ "srEpisodes": 200,
1466
+ "lscr": 25.0,
1467
+ "lscrScope": "four-channel-mean",
1468
+ "lscrEpisodes": 200,
1469
+ "efficiency": 1.1727624145108535,
1470
+ "efficiencyStatus": "reported"
1471
+ }
1472
+ },
1473
+ "79": {
1474
+ "channels": {
1475
+ "None": {
1476
+ "status": "reported",
1477
+ "episodes": 50,
1478
+ "successes": 42,
1479
+ "lscr": 84.0,
1480
+ "source": {
1481
+ "runId": "pyrua-dex_r01_79_piano_player-562b0d928f",
1482
+ "perEpisodeFile": "per_task_eval/79_bimanual_piano_melody/None/per_episode.jsonl",
1483
+ "summaryFile": "per_task_eval/79_bimanual_piano_melody/None/summary.json"
1484
+ }
1485
+ },
1486
+ "Equi.": {
1487
+ "status": "reported",
1488
+ "episodes": 50,
1489
+ "successes": 41,
1490
+ "lscr": 82.0,
1491
+ "source": {
1492
+ "runId": "pyrua-dex_r01_79_piano_player-562b0d928f",
1493
+ "perEpisodeFile": "per_task_eval/79_bimanual_piano_melody/Equi/per_episode.jsonl",
1494
+ "summaryFile": "per_task_eval/79_bimanual_piano_melody/Equi/summary.json"
1495
+ }
1496
+ },
1497
+ "Inv.": {
1498
+ "status": "reported",
1499
+ "episodes": 50,
1500
+ "successes": 43,
1501
+ "lscr": 86.0,
1502
+ "source": {
1503
+ "runId": "pyrua-dex_r01_79_piano_player-562b0d928f",
1504
+ "perEpisodeFile": "per_task_eval/79_bimanual_piano_melody/Inv/per_episode.jsonl",
1505
+ "summaryFile": "per_task_eval/79_bimanual_piano_melody/Inv/summary.json"
1506
+ }
1507
+ },
1508
+ "Full": {
1509
+ "status": "reported",
1510
+ "episodes": 50,
1511
+ "successes": 40,
1512
+ "lscr": 80.0,
1513
+ "source": {
1514
+ "runId": "pyrua-dex_r01_79_piano_player-562b0d928f",
1515
+ "perEpisodeFile": "per_task_eval/79_bimanual_piano_melody/Full/per_episode.jsonl",
1516
+ "summaryFile": "per_task_eval/79_bimanual_piano_melody/Full/summary.json"
1517
+ }
1518
+ }
1519
+ },
1520
+ "mean": {
1521
+ "status": "reported",
1522
+ "sr": 83.0,
1523
+ "srScope": "four-channel-mean",
1524
+ "srEpisodes": 200,
1525
+ "lscr": 83.0,
1526
+ "lscrScope": "four-channel-mean",
1527
+ "lscrEpisodes": 200,
1528
+ "efficiency": 1.1754075860921644,
1529
+ "efficiencyStatus": "reported"
1530
+ }
1531
+ },
1532
+ "80": {
1533
+ "channels": {
1534
+ "None": {
1535
+ "status": "reported",
1536
+ "episodes": 50,
1537
+ "successes": 47,
1538
+ "lscr": 95.33333333333334,
1539
+ "source": {
1540
+ "runId": "pyrua-dex_r01_80_desk_final-3302cc8020",
1541
+ "perEpisodeFile": "per_task_eval/80_gaming_desk_setup/None/per_episode.jsonl",
1542
+ "summaryFile": "per_task_eval/80_gaming_desk_setup/None/summary.json"
1543
+ }
1544
+ },
1545
+ "Equi.": {
1546
+ "status": "reported",
1547
+ "episodes": 50,
1548
+ "successes": 41,
1549
+ "lscr": 85.33333333333333,
1550
+ "source": {
1551
+ "runId": "pyrua-dex_r01_80_desk_final-3302cc8020",
1552
+ "perEpisodeFile": "per_task_eval/80_gaming_desk_setup/Equi/per_episode.jsonl",
1553
+ "summaryFile": "per_task_eval/80_gaming_desk_setup/Equi/summary.json"
1554
+ }
1555
+ },
1556
+ "Inv.": {
1557
+ "status": "reported",
1558
+ "episodes": 50,
1559
+ "successes": 41,
1560
+ "lscr": 86.66666666666667,
1561
+ "source": {
1562
+ "runId": "pyrua-dex_r01_80_desk_final-3302cc8020",
1563
+ "perEpisodeFile": "per_task_eval/80_gaming_desk_setup/Inv/per_episode.jsonl",
1564
+ "summaryFile": "per_task_eval/80_gaming_desk_setup/Inv/summary.json"
1565
+ }
1566
+ },
1567
+ "Full": {
1568
+ "status": "reported",
1569
+ "episodes": 50,
1570
+ "successes": 38,
1571
+ "lscr": 80.0,
1572
+ "source": {
1573
+ "runId": "pyrua-dex_r01_80_desk_final-3302cc8020",
1574
+ "perEpisodeFile": "per_task_eval/80_gaming_desk_setup/Full/per_episode.jsonl",
1575
+ "summaryFile": "per_task_eval/80_gaming_desk_setup/Full/summary.json"
1576
+ }
1577
+ }
1578
+ },
1579
+ "mean": {
1580
+ "status": "reported",
1581
+ "sr": 83.5,
1582
+ "srScope": "four-channel-mean",
1583
+ "srEpisodes": 200,
1584
+ "lscr": 86.83333333333334,
1585
+ "lscrScope": "four-channel-mean",
1586
+ "lscrEpisodes": 200,
1587
+ "efficiency": 0.8471351212516856,
1588
+ "efficiencyStatus": "reported"
1589
+ }
1590
+ }
1591
+ },
1592
+ "evaluation": {
1593
+ "schemaVersion": "1.0",
1594
+ "isTemplate": false,
1595
+ "protocol": {
1596
+ "benchmarkCommit": "ae870fb0e1f6237f20c8302198ee3850118892e1",
1597
+ "taskCount": 26,
1598
+ "channels": [
1599
+ "None",
1600
+ "Equi.",
1601
+ "Inv.",
1602
+ "Full"
1603
+ ],
1604
+ "episodesPerChannel": 50,
1605
+ "seeds": {
1606
+ "03": {
1607
+ "None": {
1608
+ "base": 100300000,
1609
+ "first": 1,
1610
+ "last": 50
1611
+ },
1612
+ "Equi.": {
1613
+ "base": 100300000,
1614
+ "first": 1,
1615
+ "last": 50
1616
+ },
1617
+ "Inv.": {
1618
+ "base": 100300000,
1619
+ "first": 1,
1620
+ "last": 50
1621
+ },
1622
+ "Full": {
1623
+ "base": 200300000,
1624
+ "first": 1,
1625
+ "last": 50
1626
+ }
1627
+ },
1628
+ "06": {
1629
+ "None": {
1630
+ "base": 100600000,
1631
+ "first": 1,
1632
+ "last": 50
1633
+ },
1634
+ "Equi.": {
1635
+ "base": 100600000,
1636
+ "first": 1,
1637
+ "last": 50
1638
+ },
1639
+ "Inv.": {
1640
+ "base": 100600000,
1641
+ "first": 1,
1642
+ "last": 50
1643
+ },
1644
+ "Full": {
1645
+ "base": 200600000,
1646
+ "first": 1,
1647
+ "last": 50
1648
+ }
1649
+ },
1650
+ "07": {
1651
+ "None": {
1652
+ "base": 100700000,
1653
+ "first": 1,
1654
+ "last": 50
1655
+ },
1656
+ "Equi.": {
1657
+ "base": 100700000,
1658
+ "first": 1,
1659
+ "last": 50
1660
+ },
1661
+ "Inv.": {
1662
+ "base": 100700000,
1663
+ "first": 1,
1664
+ "last": 50
1665
+ },
1666
+ "Full": {
1667
+ "base": 200700000,
1668
+ "first": 1,
1669
+ "last": 50
1670
+ }
1671
+ },
1672
+ "08": {
1673
+ "None": {
1674
+ "base": 100800000,
1675
+ "first": 1,
1676
+ "last": 50
1677
+ },
1678
+ "Equi.": {
1679
+ "base": 100800000,
1680
+ "first": 1,
1681
+ "last": 50
1682
+ },
1683
+ "Inv.": {
1684
+ "base": 100800000,
1685
+ "first": 1,
1686
+ "last": 50
1687
+ },
1688
+ "Full": {
1689
+ "base": 200800000,
1690
+ "first": 1,
1691
+ "last": 50
1692
+ }
1693
+ },
1694
+ "09": {
1695
+ "None": {
1696
+ "base": 100900000,
1697
+ "first": 1,
1698
+ "last": 50
1699
+ },
1700
+ "Equi.": {
1701
+ "base": 100900000,
1702
+ "first": 1,
1703
+ "last": 50
1704
+ },
1705
+ "Inv.": {
1706
+ "base": 100900000,
1707
+ "first": 1,
1708
+ "last": 50
1709
+ },
1710
+ "Full": {
1711
+ "base": 200900000,
1712
+ "first": 1,
1713
+ "last": 50
1714
+ }
1715
+ },
1716
+ "12": {
1717
+ "None": {
1718
+ "base": 101200000,
1719
+ "first": 1,
1720
+ "last": 50
1721
+ },
1722
+ "Equi.": {
1723
+ "base": 101200000,
1724
+ "first": 1,
1725
+ "last": 50
1726
+ },
1727
+ "Inv.": {
1728
+ "base": 101200000,
1729
+ "first": 1,
1730
+ "last": 50
1731
+ },
1732
+ "Full": {
1733
+ "base": 201200000,
1734
+ "first": 1,
1735
+ "last": 50
1736
+ }
1737
+ },
1738
+ "21": {
1739
+ "None": {
1740
+ "base": 102100000,
1741
+ "first": 1,
1742
+ "last": 50
1743
+ },
1744
+ "Equi.": {
1745
+ "base": 102100000,
1746
+ "first": 1,
1747
+ "last": 50
1748
+ },
1749
+ "Inv.": {
1750
+ "base": 102100000,
1751
+ "first": 1,
1752
+ "last": 50
1753
+ },
1754
+ "Full": {
1755
+ "base": 202100000,
1756
+ "first": 1,
1757
+ "last": 50
1758
+ }
1759
+ },
1760
+ "22": {
1761
+ "None": {
1762
+ "base": 102200000,
1763
+ "first": 1,
1764
+ "last": 50
1765
+ },
1766
+ "Equi.": {
1767
+ "base": 102200000,
1768
+ "first": 1,
1769
+ "last": 50
1770
+ },
1771
+ "Inv.": {
1772
+ "base": 102200000,
1773
+ "first": 1,
1774
+ "last": 50
1775
+ },
1776
+ "Full": {
1777
+ "base": 202200000,
1778
+ "first": 1,
1779
+ "last": 50
1780
+ }
1781
+ },
1782
+ "24": {
1783
+ "None": {
1784
+ "base": 102400000,
1785
+ "first": 1,
1786
+ "last": 50
1787
+ },
1788
+ "Equi.": {
1789
+ "base": 102400000,
1790
+ "first": 1,
1791
+ "last": 50
1792
+ },
1793
+ "Inv.": {
1794
+ "base": 102400000,
1795
+ "first": 1,
1796
+ "last": 50
1797
+ },
1798
+ "Full": {
1799
+ "base": 202400000,
1800
+ "first": 1,
1801
+ "last": 50
1802
+ }
1803
+ },
1804
+ "26": {
1805
+ "None": {
1806
+ "base": 102600000,
1807
+ "first": 1,
1808
+ "last": 50
1809
+ },
1810
+ "Equi.": {
1811
+ "base": 102600000,
1812
+ "first": 1,
1813
+ "last": 50
1814
+ },
1815
+ "Inv.": {
1816
+ "base": 102600000,
1817
+ "first": 1,
1818
+ "last": 50
1819
+ },
1820
+ "Full": {
1821
+ "base": 202600000,
1822
+ "first": 1,
1823
+ "last": 50
1824
+ }
1825
+ },
1826
+ "27": {
1827
+ "None": {
1828
+ "base": 102700000,
1829
+ "first": 1,
1830
+ "last": 50
1831
+ },
1832
+ "Equi.": {
1833
+ "base": 102700000,
1834
+ "first": 1,
1835
+ "last": 50
1836
+ },
1837
+ "Inv.": {
1838
+ "base": 102700000,
1839
+ "first": 1,
1840
+ "last": 50
1841
+ },
1842
+ "Full": {
1843
+ "base": 202700000,
1844
+ "first": 1,
1845
+ "last": 50
1846
+ }
1847
+ },
1848
+ "32": {
1849
+ "None": {
1850
+ "base": 103200000,
1851
+ "first": 1,
1852
+ "last": 50
1853
+ },
1854
+ "Equi.": {
1855
+ "base": 103200000,
1856
+ "first": 1,
1857
+ "last": 50
1858
+ },
1859
+ "Inv.": {
1860
+ "base": 103200000,
1861
+ "first": 1,
1862
+ "last": 50
1863
+ },
1864
+ "Full": {
1865
+ "base": 203200000,
1866
+ "first": 1,
1867
+ "last": 50
1868
+ }
1869
+ },
1870
+ "34": {
1871
+ "None": {
1872
+ "base": 103400000,
1873
+ "first": 1,
1874
+ "last": 50
1875
+ },
1876
+ "Equi.": {
1877
+ "base": 103400000,
1878
+ "first": 1,
1879
+ "last": 50
1880
+ },
1881
+ "Inv.": {
1882
+ "base": 103400000,
1883
+ "first": 1,
1884
+ "last": 50
1885
+ },
1886
+ "Full": {
1887
+ "base": 203400000,
1888
+ "first": 1,
1889
+ "last": 50
1890
+ }
1891
+ },
1892
+ "42": {
1893
+ "None": {
1894
+ "base": 104200000,
1895
+ "first": 1,
1896
+ "last": 50
1897
+ },
1898
+ "Equi.": {
1899
+ "base": 104200000,
1900
+ "first": 1,
1901
+ "last": 50
1902
+ },
1903
+ "Inv.": {
1904
+ "base": 104200000,
1905
+ "first": 1,
1906
+ "last": 50
1907
+ },
1908
+ "Full": {
1909
+ "base": 204200000,
1910
+ "first": 1,
1911
+ "last": 50
1912
+ }
1913
+ },
1914
+ "43": {
1915
+ "None": {
1916
+ "base": 104300000,
1917
+ "first": 1,
1918
+ "last": 50
1919
+ },
1920
+ "Equi.": {
1921
+ "base": 104300000,
1922
+ "first": 1,
1923
+ "last": 50
1924
+ },
1925
+ "Inv.": {
1926
+ "base": 104300000,
1927
+ "first": 1,
1928
+ "last": 50
1929
+ },
1930
+ "Full": {
1931
+ "base": 204300000,
1932
+ "first": 1,
1933
+ "last": 50
1934
+ }
1935
+ },
1936
+ "44": {
1937
+ "None": {
1938
+ "base": 104400000,
1939
+ "first": 1,
1940
+ "last": 50
1941
+ },
1942
+ "Equi.": {
1943
+ "base": 104400000,
1944
+ "first": 1,
1945
+ "last": 50
1946
+ },
1947
+ "Inv.": {
1948
+ "base": 104400000,
1949
+ "first": 1,
1950
+ "last": 50
1951
+ },
1952
+ "Full": {
1953
+ "base": 204400000,
1954
+ "first": 1,
1955
+ "last": 50
1956
+ }
1957
+ },
1958
+ "51": {
1959
+ "None": {
1960
+ "base": 105100000,
1961
+ "first": 1,
1962
+ "last": 50
1963
+ },
1964
+ "Equi.": {
1965
+ "base": 105100000,
1966
+ "first": 1,
1967
+ "last": 50
1968
+ },
1969
+ "Inv.": {
1970
+ "base": 105100000,
1971
+ "first": 1,
1972
+ "last": 50
1973
+ },
1974
+ "Full": {
1975
+ "base": 205100000,
1976
+ "first": 1,
1977
+ "last": 50
1978
+ }
1979
+ },
1980
+ "60": {
1981
+ "None": {
1982
+ "base": 106000000,
1983
+ "first": 1,
1984
+ "last": 50
1985
+ },
1986
+ "Equi.": {
1987
+ "base": 106000000,
1988
+ "first": 1,
1989
+ "last": 50
1990
+ },
1991
+ "Inv.": {
1992
+ "base": 106000000,
1993
+ "first": 1,
1994
+ "last": 50
1995
+ },
1996
+ "Full": {
1997
+ "base": 206000000,
1998
+ "first": 1,
1999
+ "last": 50
2000
+ }
2001
+ },
2002
+ "61": {
2003
+ "None": {
2004
+ "base": 106100000,
2005
+ "first": 1,
2006
+ "last": 50
2007
+ },
2008
+ "Equi.": {
2009
+ "base": 106100000,
2010
+ "first": 1,
2011
+ "last": 50
2012
+ },
2013
+ "Inv.": {
2014
+ "base": 106100000,
2015
+ "first": 1,
2016
+ "last": 50
2017
+ },
2018
+ "Full": {
2019
+ "base": 206100000,
2020
+ "first": 1,
2021
+ "last": 50
2022
+ }
2023
+ },
2024
+ "62": {
2025
+ "None": {
2026
+ "base": 106200000,
2027
+ "first": 1,
2028
+ "last": 50
2029
+ },
2030
+ "Equi.": {
2031
+ "base": 106200000,
2032
+ "first": 1,
2033
+ "last": 50
2034
+ },
2035
+ "Inv.": {
2036
+ "base": 106200000,
2037
+ "first": 1,
2038
+ "last": 50
2039
+ },
2040
+ "Full": {
2041
+ "base": 206200000,
2042
+ "first": 1,
2043
+ "last": 50
2044
+ }
2045
+ },
2046
+ "64": {
2047
+ "None": {
2048
+ "base": 106400000,
2049
+ "first": 1,
2050
+ "last": 50
2051
+ },
2052
+ "Equi.": {
2053
+ "base": 106400000,
2054
+ "first": 1,
2055
+ "last": 50
2056
+ },
2057
+ "Inv.": {
2058
+ "base": 106400000,
2059
+ "first": 1,
2060
+ "last": 50
2061
+ },
2062
+ "Full": {
2063
+ "base": 206400000,
2064
+ "first": 1,
2065
+ "last": 50
2066
+ }
2067
+ },
2068
+ "67": {
2069
+ "None": {
2070
+ "base": 106700000,
2071
+ "first": 1,
2072
+ "last": 50
2073
+ },
2074
+ "Equi.": {
2075
+ "base": 106700000,
2076
+ "first": 1,
2077
+ "last": 50
2078
+ },
2079
+ "Inv.": {
2080
+ "base": 106700000,
2081
+ "first": 1,
2082
+ "last": 50
2083
+ },
2084
+ "Full": {
2085
+ "base": 206700000,
2086
+ "first": 1,
2087
+ "last": 50
2088
+ }
2089
+ },
2090
+ "73": {
2091
+ "None": {
2092
+ "base": 107300000,
2093
+ "first": 1,
2094
+ "last": 50
2095
+ },
2096
+ "Equi.": {
2097
+ "base": 107300000,
2098
+ "first": 1,
2099
+ "last": 50
2100
+ },
2101
+ "Inv.": {
2102
+ "base": 107300000,
2103
+ "first": 1,
2104
+ "last": 50
2105
+ },
2106
+ "Full": {
2107
+ "base": 207300000,
2108
+ "first": 1,
2109
+ "last": 50
2110
+ }
2111
+ },
2112
+ "76": {
2113
+ "None": {
2114
+ "base": 107600000,
2115
+ "first": 1,
2116
+ "last": 50
2117
+ },
2118
+ "Equi.": {
2119
+ "base": 107600000,
2120
+ "first": 1,
2121
+ "last": 50
2122
+ },
2123
+ "Inv.": {
2124
+ "base": 107600000,
2125
+ "first": 1,
2126
+ "last": 50
2127
+ },
2128
+ "Full": {
2129
+ "base": 207600000,
2130
+ "first": 1,
2131
+ "last": 50
2132
+ }
2133
+ },
2134
+ "79": {
2135
+ "None": {
2136
+ "base": 107900000,
2137
+ "first": 1,
2138
+ "last": 50
2139
+ },
2140
+ "Equi.": {
2141
+ "base": 107900000,
2142
+ "first": 1,
2143
+ "last": 50
2144
+ },
2145
+ "Inv.": {
2146
+ "base": 107900000,
2147
+ "first": 1,
2148
+ "last": 50
2149
+ },
2150
+ "Full": {
2151
+ "base": 207900000,
2152
+ "first": 1,
2153
+ "last": 50
2154
+ }
2155
+ },
2156
+ "80": {
2157
+ "None": {
2158
+ "base": 108000000,
2159
+ "first": 1,
2160
+ "last": 50
2161
+ },
2162
+ "Equi.": {
2163
+ "base": 108000000,
2164
+ "first": 1,
2165
+ "last": 50
2166
+ },
2167
+ "Inv.": {
2168
+ "base": 108000000,
2169
+ "first": 1,
2170
+ "last": 50
2171
+ },
2172
+ "Full": {
2173
+ "base": 208000000,
2174
+ "first": 1,
2175
+ "last": 50
2176
+ }
2177
+ }
2178
+ },
2179
+ "stepUnit": "physics_steps",
2180
+ "physicsHz": 60,
2181
+ "controlHz": 20,
2182
+ "evaluationProtocol": "reach_and_stop",
2183
+ "earlyStop": true,
2184
+ "executionStepBudget": {
2185
+ "rule": "3 * ceil(1.5 * expert_time_step / 3) physics steps (the benchmark's own episode limit, 1x)",
2186
+ "perTaskPhysicsSteps": {
2187
+ "03": 3246,
2188
+ "06": 1920,
2189
+ "07": 1806,
2190
+ "08": 3432,
2191
+ "09": 1926,
2192
+ "12": 3447,
2193
+ "21": 2481,
2194
+ "22": 2043,
2195
+ "24": 4626,
2196
+ "26": 2613,
2197
+ "27": 3570,
2198
+ "32": 2556,
2199
+ "34": 3240,
2200
+ "42": 2004,
2201
+ "43": 4125,
2202
+ "44": 3183,
2203
+ "51": 3069,
2204
+ "60": 2985,
2205
+ "61": 3102,
2206
+ "62": 3231,
2207
+ "64": 1995,
2208
+ "67": 2892,
2209
+ "73": 3639,
2210
+ "76": 2958,
2211
+ "79": 2043,
2212
+ "80": 1779
2213
+ },
2214
+ "perTaskPolicySteps": {
2215
+ "03": 1082,
2216
+ "06": 640,
2217
+ "07": 602,
2218
+ "08": 1144,
2219
+ "09": 642,
2220
+ "12": 1149,
2221
+ "21": 827,
2222
+ "22": 681,
2223
+ "24": 1542,
2224
+ "26": 871,
2225
+ "27": 1190,
2226
+ "32": 852,
2227
+ "34": 1080,
2228
+ "42": 668,
2229
+ "43": 1375,
2230
+ "44": 1061,
2231
+ "51": 1023,
2232
+ "60": 995,
2233
+ "61": 1034,
2234
+ "62": 1077,
2235
+ "64": 665,
2236
+ "67": 964,
2237
+ "73": 1213,
2238
+ "76": 986,
2239
+ "79": 681,
2240
+ "80": 593
2241
+ }
2242
+ },
2243
+ "dwellTimeSecondsByTask": {
2244
+ "03": 0.5,
2245
+ "06": 0.5,
2246
+ "07": 0.5,
2247
+ "08": 0.5,
2248
+ "09": 0.5,
2249
+ "12": 0.5,
2250
+ "21": 0.5,
2251
+ "22": 0.5,
2252
+ "24": 0.5,
2253
+ "26": 0.5,
2254
+ "27": 0.5,
2255
+ "32": 0.5,
2256
+ "34": 0.5,
2257
+ "42": 0.5,
2258
+ "43": 0.5,
2259
+ "44": 0.5,
2260
+ "51": 0.5,
2261
+ "60": 0.5,
2262
+ "61": 0.5,
2263
+ "62": 0.5,
2264
+ "64": 0.5,
2265
+ "67": 3.0,
2266
+ "73": 0.5,
2267
+ "76": 0.5,
2268
+ "79": 0.5,
2269
+ "80": 0.5
2270
+ },
2271
+ "expertTimeStepByTask": {
2272
+ "03": 2164,
2273
+ "06": 1280,
2274
+ "07": 1203,
2275
+ "08": 2288,
2276
+ "09": 1283,
2277
+ "12": 2297,
2278
+ "21": 1654,
2279
+ "22": 1361,
2280
+ "24": 3083,
2281
+ "26": 1742,
2282
+ "27": 2380,
2283
+ "32": 1703,
2284
+ "34": 2160,
2285
+ "42": 1336,
2286
+ "43": 2750,
2287
+ "44": 2122,
2288
+ "51": 2046,
2289
+ "60": 1989,
2290
+ "61": 2067,
2291
+ "62": 2154,
2292
+ "64": 1329,
2293
+ "67": 1928,
2294
+ "73": 2426,
2295
+ "76": 1972,
2296
+ "79": 1362,
2297
+ "80": 1185
2298
+ },
2299
+ "expertReference": "paper-expert-reference (expert_time_step of the scene files at the benchmark commit)",
2300
+ "protocolChanges": [
2301
+ "Method: one frozen program per task, written and tested by GPT-6 Astra coding agents on development layouts (seeds disjoint from the evaluation seeds, with the simulator's stage feedback). No human demonstration data, no trained weights, no language-model calls during evaluation.",
2302
+ "Task description: the benchmark's instruction together with a goal text saying in words what completing the task means.",
2303
+ "Inputs at evaluation: besides RGB cameras, joint positions and tactile, the program uses the episode's camera calibration, forward kinematics and a segmentation model (SAM3); no simulator state or success/stage signals."
2304
+ ]
2305
+ },
2306
+ "coverage": {
2307
+ "evaluatedTaskCount": 26,
2308
+ "evaluatedEpisodeCount": 5200,
2309
+ "completedTaskCount": {
2310
+ "None": 26,
2311
+ "Equi.": 26,
2312
+ "Inv.": 26,
2313
+ "Full": 26,
2314
+ "mean": 26
2315
+ },
2316
+ "expectedFullEpisodeCount": 1300,
2317
+ "expectedFourChannelEpisodeCount": 5200,
2318
+ "missingTaskChannels": {}
2319
+ },
2320
+ "aggregation": {
2321
+ "units": {
2322
+ "sr": "percent",
2323
+ "lscr": "percent",
2324
+ "efficiency": "ratio",
2325
+ "steps": "physics_steps"
2326
+ },
2327
+ "sr": "100 * successes / episodes per task/channel; equal channel mean per task; equal task mean over all 26 settings.",
2328
+ "lscr": "100 * mean episode latched_stage_completion_rate, including failures; equal channel mean per task; equal task mean over all 26 settings.",
2329
+ "efficiency": "expert_time_step / mean successful steps_to_stable_success. Pool successful episodes across all four channels before calculating task mean efficiency. Overall values average task efficiencies over tasks with success; report valid-task counts.",
2330
+ "missing": "null means unreported (NR). Efficiency without successes is null with no-success status. Valid-task counts are null when unreported, and 0 for an evaluated scope without successes.",
2331
+ "scope": "Overall channel aggregates require all 26 settings. Four-channel means require all four channels. Never substitute Full-only scores for means or include duplicate reruns."
2332
+ },
2333
+ "logs": {
2334
+ "repository": "Bench2Dex/Leaderboard",
2335
+ "policyPath": "pyrua-dex",
2336
+ "revision": null,
2337
+ "status": "not-archived"
2338
+ },
2339
+ "selectedPrograms": {
2340
+ "03": {
2341
+ "program": "wine_fast-e9124b6af8",
2342
+ "devComplete": "22/32"
2343
+ },
2344
+ "06": {
2345
+ "program": "fruit_controller-429ffea065",
2346
+ "devComplete": "17/32"
2347
+ },
2348
+ "07": {
2349
+ "program": "citrus_policy-3695867c40",
2350
+ "devComplete": "20/32"
2351
+ },
2352
+ "08": {
2353
+ "program": "soy_fixed-cdfd94bffc",
2354
+ "devComplete": "0/32"
2355
+ },
2356
+ "09": {
2357
+ "program": "cleaner55-5f88a39d99",
2358
+ "devComplete": "16/32"
2359
+ },
2360
+ "12": {
2361
+ "program": "measured_tools-adce3bd7ad",
2362
+ "devComplete": "0/32"
2363
+ },
2364
+ "21": {
2365
+ "program": "round_fallback-33066e1d92",
2366
+ "devComplete": "5/32"
2367
+ },
2368
+ "22": {
2369
+ "program": "toolbox-59184eb3a0",
2370
+ "devComplete": "15/32"
2371
+ },
2372
+ "24": {
2373
+ "program": "stationery_full-fc6bd676a0",
2374
+ "devComplete": "10/32"
2375
+ },
2376
+ "26": {
2377
+ "program": "tray_round-c2ad9643b5",
2378
+ "devComplete": "12/32"
2379
+ },
2380
+ "27": {
2381
+ "program": "ring_prefix_thumb_release-f1ac959cf9",
2382
+ "devComplete": "21/32"
2383
+ },
2384
+ "32": {
2385
+ "program": "tray_hybrid-c636840313",
2386
+ "devComplete": "10/32"
2387
+ },
2388
+ "34": {
2389
+ "program": "retensioned_service-3cc32133ca",
2390
+ "devComplete": "0/32"
2391
+ },
2392
+ "42": {
2393
+ "program": "trash-f39e0082b0",
2394
+ "devComplete": "10/32"
2395
+ },
2396
+ "43": {
2397
+ "program": "fridge_round-c4e0b828aa",
2398
+ "devComplete": "5/32"
2399
+ },
2400
+ "44": {
2401
+ "program": "reliable-a666046e78",
2402
+ "devComplete": "19/32"
2403
+ },
2404
+ "51": {
2405
+ "program": "toilet_final-4a83db3920",
2406
+ "devComplete": "28/32"
2407
+ },
2408
+ "60": {
2409
+ "program": "meal-560f17c29a",
2410
+ "devComplete": "15/32"
2411
+ },
2412
+ "61": {
2413
+ "program": "medicine_point_v2-06e06faffc",
2414
+ "devComplete": "27/32"
2415
+ },
2416
+ "62": {
2417
+ "program": "shoebox-8b45ce4db8",
2418
+ "devComplete": "14/32"
2419
+ },
2420
+ "64": {
2421
+ "program": "ball_sort-7a836c8a10",
2422
+ "devComplete": "19/32"
2423
+ },
2424
+ "67": {
2425
+ "program": "faucet_recovery-ee8f8b886b",
2426
+ "devComplete": "17/32"
2427
+ },
2428
+ "73": {
2429
+ "program": "yellowbounds-df764f33e4",
2430
+ "devComplete": "0/32"
2431
+ },
2432
+ "76": {
2433
+ "program": "serving_palm-139449a835",
2434
+ "devComplete": "5/32"
2435
+ },
2436
+ "79": {
2437
+ "program": "piano_player-562b0d928f",
2438
+ "devComplete": "27/32"
2439
+ },
2440
+ "80": {
2441
+ "program": "desk_final-3302cc8020",
2442
+ "devComplete": "29/32"
2443
+ }
2444
+ }
2445
+ }
2446
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/native_shards/000/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/native_shards/000/summary.json ADDED
@@ -0,0 +1,193 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 100300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 18,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.72,
50
+ "success_rate_ci": {
51
+ "mean": 0.72,
52
+ "ci_low": 0.52,
53
+ "ci_high": 0.88
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.8666666666666667
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 34.758333333332644
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.72,
64
+ "hard_violation_rate": 0.08,
65
+ "drop_rate": 0.08,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 3.3188961351454505e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.72,
74
+ "at_end_success_rate": 0.72,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.72
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.84,
81
+ "current_chain_depth_progress_score": 0.84,
82
+ "latched_chain_depth_progress_score": 0.8666666666666667,
83
+ "avg_current_chain_depth": 2.52,
84
+ "avg_latched_chain_depth": 2.6
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.041409037163462,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2085.5,
91
+ "avg_policy_steps_to_success": 695.4444444444445,
92
+ "avg_policy_queries_to_success": 695.4444444444445
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 3.3188961351454505e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_1": 1,
98
+ "drop_to_table_obj_194_wineglass_3": 1
99
+ },
100
+ "robot_constraint_violation_rate": 0.0,
101
+ "finger_joint_limit_saturation_rate": 0.0,
102
+ "joint_limit_active_step_rate": 0.0,
103
+ "finger_joint_limit_saturation_step_rate": 0.0,
104
+ "max_joint_limit_excess_rad": 0.004140238277614117,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {},
107
+ "joint_limit_hard_counts": {},
108
+ "joint_limit_hard_episode_events": 0,
109
+ "finger_joint_limit_saturation_counts": {},
110
+ "joint_limit_max_excess_by_joint": {
111
+ "index_q1": 5.268533387337015e-10,
112
+ "index_q2": 7.266733326183328e-10,
113
+ "multi_middle_q1": 5.784106527073618e-10,
114
+ "multi_pinky_q1": 0.0018935843836516142,
115
+ "multi_pinky_q2": 0.0030601536855101585,
116
+ "multi_ring_q1": 0.0018982011824846268,
117
+ "multi_ring_q2": 0.004140238277614117,
118
+ "multi_thumb_q1": 1.1553316081602816e-09,
119
+ "multi_thumb_q2": 1.3309879587364293e-10,
120
+ "pinky_q1": 0.0004376902070362121,
121
+ "pinky_q2": 0.0030433775391429663,
122
+ "ring_q1": 0.0009009296190924942,
123
+ "ring_q2": 0.004095674958080053,
124
+ "thumb_q1": 7.678177538217312e-10,
125
+ "thumb_q2": 3.740576459687617e-10
126
+ },
127
+ "coverage": {
128
+ "hard_limit_source": {
129
+ "available": 25,
130
+ "unavailable": 0,
131
+ "unknown": 0
132
+ },
133
+ "joint_role_manifest": {
134
+ "available": 25,
135
+ "unavailable": 0,
136
+ "unknown": 0
137
+ }
138
+ },
139
+ "hard_limit_sources": {
140
+ "runtime.joint_pos_limits": 25
141
+ },
142
+ "soft_limit_sources": {
143
+ "runtime.soft_joint_pos_limits": 25
144
+ },
145
+ "unknown_runtime_joint_names": [],
146
+ "baseline_checks": {
147
+ "post_home": {
148
+ "checked": 25,
149
+ "hard_violation_episodes": 0
150
+ },
151
+ "post_reset": {
152
+ "checked": 25,
153
+ "hard_violation_episodes": 0
154
+ }
155
+ }
156
+ }
157
+ },
158
+ "robot_motion": {
159
+ "joint_acceleration_rms": 7.191288926042113,
160
+ "joint_effort_rms": 83.84970210528189,
161
+ "joint_jerk_rms": 732.1593030597073,
162
+ "joint_velocity_rms": 0.4414949336875708
163
+ },
164
+ "grasp": {
165
+ "kinematic_grasp_stability_index": 0.3347057196267719,
166
+ "mean_kinematic_grasp_stability": 0.10816516510162,
167
+ "per_object_best_kinematic_grasp_stability": {
168
+ "obj_194_wineglass_1": 0.14110855645154793,
169
+ "obj_194_wineglass_2": 0.12242809936604548,
170
+ "obj_194_wineglass_3": 0.2298768863834962
171
+ },
172
+ "grasp_gsi_diagnostics": {
173
+ "mean_anti_slip_score": 0.9401418870144213,
174
+ "mean_duration_score": 0.12034767314296647,
175
+ "mean_hold_score": 0.0688174583857305,
176
+ "mean_kinematic_stability_score": 0.03682100956517138,
177
+ "mean_legacy_gsi": 0.26411651873446346,
178
+ "mean_lift_score": 0.17504900281006489,
179
+ "mean_motion_stability_score": 0.07270029117154586
180
+ }
181
+ },
182
+ "tool": {
183
+ "tool_selection_accuracy": null,
184
+ "tool_switch_success_rate": null,
185
+ "tool_switch_total": 0,
186
+ "tool_switch_successful": 0
187
+ },
188
+ "runtime": {
189
+ "avg_steps": 2410.44,
190
+ "error_count": 0
191
+ }
192
+ }
193
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/native_shards/001/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/native_shards/001/summary.json ADDED
@@ -0,0 +1,187 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 100300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 20,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.8,
50
+ "success_rate_ci": {
51
+ "mean": 0.8,
52
+ "ci_low": 0.64,
53
+ "ci_high": 0.96
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.9466666666666667
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 34.50999999999932
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.8,
64
+ "hard_violation_rate": 0.08,
65
+ "drop_rate": 0.08,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 3.469692238298463e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.8,
74
+ "at_end_success_rate": 0.8,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.8
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.9466666666666667,
81
+ "current_chain_depth_progress_score": 0.9466666666666667,
82
+ "latched_chain_depth_progress_score": 0.9466666666666667,
83
+ "avg_current_chain_depth": 2.84,
84
+ "avg_latched_chain_depth": 2.84
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.0485195973055457,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2070.6,
91
+ "avg_policy_steps_to_success": 690.5,
92
+ "avg_policy_queries_to_success": 690.5
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 3.469692238298463e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_1": 2
98
+ },
99
+ "robot_constraint_violation_rate": 0.04,
100
+ "finger_joint_limit_saturation_rate": 0.0,
101
+ "joint_limit_active_step_rate": 0.0,
102
+ "finger_joint_limit_saturation_step_rate": 0.0,
103
+ "max_joint_limit_excess_rad": 0.005060350056737661,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {},
106
+ "joint_limit_hard_counts": {
107
+ "multi_ring_q2": 1
108
+ },
109
+ "joint_limit_hard_episode_events": 1,
110
+ "finger_joint_limit_saturation_counts": {},
111
+ "joint_limit_max_excess_by_joint": {
112
+ "multi_pinky_q1": 0.0013607860310003161,
113
+ "multi_pinky_q2": 0.0036594911944121122,
114
+ "multi_ring_q1": 0.0016937669133767486,
115
+ "multi_ring_q2": 0.005060350056737661,
116
+ "pinky_q1": 0.0007309927023015916,
117
+ "pinky_q2": 0.003384663024917245,
118
+ "ring_q1": 0.0008507093298248947,
119
+ "ring_q2": 0.004674209747463465
120
+ },
121
+ "coverage": {
122
+ "hard_limit_source": {
123
+ "available": 25,
124
+ "unavailable": 0,
125
+ "unknown": 0
126
+ },
127
+ "joint_role_manifest": {
128
+ "available": 25,
129
+ "unavailable": 0,
130
+ "unknown": 0
131
+ }
132
+ },
133
+ "hard_limit_sources": {
134
+ "runtime.joint_pos_limits": 25
135
+ },
136
+ "soft_limit_sources": {
137
+ "runtime.soft_joint_pos_limits": 25
138
+ },
139
+ "unknown_runtime_joint_names": [],
140
+ "baseline_checks": {
141
+ "post_home": {
142
+ "checked": 25,
143
+ "hard_violation_episodes": 0
144
+ },
145
+ "post_reset": {
146
+ "checked": 25,
147
+ "hard_violation_episodes": 0
148
+ }
149
+ }
150
+ }
151
+ },
152
+ "robot_motion": {
153
+ "joint_acceleration_rms": 7.697518726164713,
154
+ "joint_effort_rms": 84.0922219671082,
155
+ "joint_jerk_rms": 783.267105480495,
156
+ "joint_velocity_rms": 0.4876729954008316
157
+ },
158
+ "grasp": {
159
+ "kinematic_grasp_stability_index": 0.3716021417524509,
160
+ "mean_kinematic_grasp_stability": 0.16670196387470979,
161
+ "per_object_best_kinematic_grasp_stability": {
162
+ "obj_194_wineglass_1": 0.1800945941785067,
163
+ "obj_194_wineglass_2": 0.18000936751802132,
164
+ "obj_194_wineglass_3": 0.1979367727145305
165
+ },
166
+ "grasp_gsi_diagnostics": {
167
+ "mean_anti_slip_score": 0.9689389335674128,
168
+ "mean_duration_score": 0.14714850221513928,
169
+ "mean_hold_score": 0.07872301312760688,
170
+ "mean_kinematic_stability_score": 0.05602958924469813,
171
+ "mean_legacy_gsi": 0.28139435921116307,
172
+ "mean_lift_score": 0.1955555757558805,
173
+ "mean_motion_stability_score": 0.08051963717755542
174
+ }
175
+ },
176
+ "tool": {
177
+ "tool_selection_accuracy": null,
178
+ "tool_switch_success_rate": null,
179
+ "tool_switch_total": 0,
180
+ "tool_switch_successful": 0
181
+ },
182
+ "runtime": {
183
+ "avg_steps": 2305.68,
184
+ "error_count": 0
185
+ }
186
+ }
187
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Equi/summary.json ADDED
@@ -0,0 +1,195 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 100300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 38,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.76,
50
+ "success_rate_ci": {
51
+ "mean": 0.76,
52
+ "ci_low": 0.64,
53
+ "ci_high": 0.88
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.9066666666666666
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 34.62763157894668
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.76,
64
+ "hard_violation_rate": 0.08,
65
+ "drop_rate": 0.08,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 3.392619356589739e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.76,
74
+ "at_end_success_rate": 0.76,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.76
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.8933333333333333,
81
+ "current_chain_depth_progress_score": 0.8933333333333333,
82
+ "latched_chain_depth_progress_score": 0.9066666666666666,
83
+ "avg_current_chain_depth": 2.68,
84
+ "avg_latched_chain_depth": 2.72
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.0451514372382429,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2077.657894736842,
91
+ "avg_policy_steps_to_success": 692.8421052631579,
92
+ "avg_policy_queries_to_success": 692.8421052631579
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 3.392619356589739e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_1": 3,
98
+ "drop_to_table_obj_194_wineglass_3": 1
99
+ },
100
+ "robot_constraint_violation_rate": 0.02,
101
+ "finger_joint_limit_saturation_rate": 0.0,
102
+ "joint_limit_active_step_rate": 0.0,
103
+ "finger_joint_limit_saturation_step_rate": 0.0,
104
+ "max_joint_limit_excess_rad": 0.005060350056737661,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {},
107
+ "joint_limit_hard_counts": {
108
+ "multi_ring_q2": 1
109
+ },
110
+ "joint_limit_hard_episode_events": 1,
111
+ "finger_joint_limit_saturation_counts": {},
112
+ "joint_limit_max_excess_by_joint": {
113
+ "index_q1": 5.268533387337015e-10,
114
+ "index_q2": 7.266733326183328e-10,
115
+ "multi_middle_q1": 5.784106527073618e-10,
116
+ "multi_pinky_q1": 0.0018935843836516142,
117
+ "multi_pinky_q2": 0.0036594911944121122,
118
+ "multi_ring_q1": 0.0018982011824846268,
119
+ "multi_ring_q2": 0.005060350056737661,
120
+ "multi_thumb_q1": 1.1553316081602816e-09,
121
+ "multi_thumb_q2": 1.3309879587364293e-10,
122
+ "pinky_q1": 0.0007309927023015916,
123
+ "pinky_q2": 0.003384663024917245,
124
+ "ring_q1": 0.0009009296190924942,
125
+ "ring_q2": 0.004674209747463465,
126
+ "thumb_q1": 7.678177538217312e-10,
127
+ "thumb_q2": 3.740576459687617e-10
128
+ },
129
+ "coverage": {
130
+ "hard_limit_source": {
131
+ "available": 50,
132
+ "unavailable": 0,
133
+ "unknown": 0
134
+ },
135
+ "joint_role_manifest": {
136
+ "available": 50,
137
+ "unavailable": 0,
138
+ "unknown": 0
139
+ }
140
+ },
141
+ "hard_limit_sources": {
142
+ "runtime.joint_pos_limits": 50
143
+ },
144
+ "soft_limit_sources": {
145
+ "runtime.soft_joint_pos_limits": 50
146
+ },
147
+ "unknown_runtime_joint_names": [],
148
+ "baseline_checks": {
149
+ "post_home": {
150
+ "checked": 50,
151
+ "hard_violation_episodes": 0
152
+ },
153
+ "post_reset": {
154
+ "checked": 50,
155
+ "hard_violation_episodes": 0
156
+ }
157
+ }
158
+ }
159
+ },
160
+ "robot_motion": {
161
+ "joint_acceleration_rms": 7.444403826103413,
162
+ "joint_effort_rms": 83.97096203619505,
163
+ "joint_jerk_rms": 757.7132042701012,
164
+ "joint_velocity_rms": 0.4645839645442012
165
+ },
166
+ "grasp": {
167
+ "kinematic_grasp_stability_index": 0.35315393068961143,
168
+ "mean_kinematic_grasp_stability": 0.1386530811292709,
169
+ "per_object_best_kinematic_grasp_stability": {
170
+ "obj_194_wineglass_1": 0.16060157531502733,
171
+ "obj_194_wineglass_2": 0.1512187334420334,
172
+ "obj_194_wineglass_3": 0.21390682954901336
173
+ },
174
+ "grasp_gsi_diagnostics": {
175
+ "mean_anti_slip_score": 0.954540410290917,
176
+ "mean_duration_score": 0.13374808767905288,
177
+ "mean_hold_score": 0.07377023575666869,
178
+ "mean_kinematic_stability_score": 0.04642529940493475,
179
+ "mean_legacy_gsi": 0.2727554389728133,
180
+ "mean_lift_score": 0.1853022892829727,
181
+ "mean_motion_stability_score": 0.07660996417455064
182
+ }
183
+ },
184
+ "tool": {
185
+ "tool_selection_accuracy": null,
186
+ "tool_switch_success_rate": null,
187
+ "tool_switch_total": 0,
188
+ "tool_switch_successful": 0
189
+ },
190
+ "runtime": {
191
+ "avg_steps": 2358.06,
192
+ "error_count": 0
193
+ }
194
+ }
195
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/native_shards/000/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/native_shards/000/summary.json ADDED
@@ -0,0 +1,185 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 200300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 20,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.8,
50
+ "success_rate_ci": {
51
+ "mean": 0.8,
52
+ "ci_low": 0.64,
53
+ "ci_high": 0.92
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.9333333333333333
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 35.28583333333261
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.8,
64
+ "hard_violation_rate": 0.04,
65
+ "drop_rate": 0.04,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 1.7072712683318253e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.84,
74
+ "at_end_success_rate": 0.8,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.8
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.9333333333333333,
81
+ "current_chain_depth_progress_score": 0.9333333333333333,
82
+ "latched_chain_depth_progress_score": 0.9333333333333333,
83
+ "avg_current_chain_depth": 2.8,
84
+ "avg_latched_chain_depth": 2.8
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.0251622739013697,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2117.15,
91
+ "avg_policy_steps_to_success": 706,
92
+ "avg_policy_queries_to_success": 706
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 1.7072712683318253e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_1": 1
98
+ },
99
+ "robot_constraint_violation_rate": 0.0,
100
+ "finger_joint_limit_saturation_rate": 0.0,
101
+ "joint_limit_active_step_rate": 0.0,
102
+ "finger_joint_limit_saturation_step_rate": 0.0,
103
+ "max_joint_limit_excess_rad": 0.0049931700341403484,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {},
106
+ "joint_limit_hard_counts": {},
107
+ "joint_limit_hard_episode_events": 0,
108
+ "finger_joint_limit_saturation_counts": {},
109
+ "joint_limit_max_excess_by_joint": {
110
+ "multi_pinky_q1": 0.0017718856688588858,
111
+ "multi_pinky_q2": 0.0035519045777618885,
112
+ "multi_ring_q1": 0.0017534798244014382,
113
+ "multi_ring_q2": 0.0049931700341403484,
114
+ "pinky_q1": 0.0005137042608112097,
115
+ "pinky_q2": 0.0025463819038122892,
116
+ "ring_q1": 0.0008729335386306047,
117
+ "ring_q2": 0.003435297403484583
118
+ },
119
+ "coverage": {
120
+ "hard_limit_source": {
121
+ "available": 25,
122
+ "unavailable": 0,
123
+ "unknown": 0
124
+ },
125
+ "joint_role_manifest": {
126
+ "available": 25,
127
+ "unavailable": 0,
128
+ "unknown": 0
129
+ }
130
+ },
131
+ "hard_limit_sources": {
132
+ "runtime.joint_pos_limits": 25
133
+ },
134
+ "soft_limit_sources": {
135
+ "runtime.soft_joint_pos_limits": 25
136
+ },
137
+ "unknown_runtime_joint_names": [],
138
+ "baseline_checks": {
139
+ "post_home": {
140
+ "checked": 25,
141
+ "hard_violation_episodes": 0
142
+ },
143
+ "post_reset": {
144
+ "checked": 25,
145
+ "hard_violation_episodes": 0
146
+ }
147
+ }
148
+ }
149
+ },
150
+ "robot_motion": {
151
+ "joint_acceleration_rms": 7.7096805453896975,
152
+ "joint_effort_rms": 83.15806395096321,
153
+ "joint_jerk_rms": 783.0863866600909,
154
+ "joint_velocity_rms": 0.4777728668990816
155
+ },
156
+ "grasp": {
157
+ "kinematic_grasp_stability_index": 0.39754329033653035,
158
+ "mean_kinematic_grasp_stability": 0.14354240111994687,
159
+ "per_object_best_kinematic_grasp_stability": {
160
+ "obj_194_wineglass_1": 0.17902588286757468,
161
+ "obj_194_wineglass_2": 0.06794326504799973,
162
+ "obj_194_wineglass_3": 0.2928101896628451
163
+ },
164
+ "grasp_gsi_diagnostics": {
165
+ "mean_anti_slip_score": 0.9480176745125708,
166
+ "mean_duration_score": 0.1558317803109365,
167
+ "mean_hold_score": 0.06454579974481048,
168
+ "mean_kinematic_stability_score": 0.03813791983699339,
169
+ "mean_legacy_gsi": 0.26700831257300306,
170
+ "mean_lift_score": 0.16473787688300598,
171
+ "mean_motion_stability_score": 0.06017505332576277
172
+ }
173
+ },
174
+ "tool": {
175
+ "tool_selection_accuracy": null,
176
+ "tool_switch_success_rate": null,
177
+ "tool_switch_total": 0,
178
+ "tool_switch_successful": 0
179
+ },
180
+ "runtime": {
181
+ "avg_steps": 2342.92,
182
+ "error_count": 0
183
+ }
184
+ }
185
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/native_shards/001/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/native_shards/001/summary.json ADDED
@@ -0,0 +1,187 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 200300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 20,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.8,
50
+ "success_rate_ci": {
51
+ "mean": 0.8,
52
+ "ci_low": 0.64,
53
+ "ci_high": 0.96
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.92
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 34.71749999999931
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.76,
64
+ "hard_violation_rate": 0.12,
65
+ "drop_rate": 0.12,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 5.182152666217547e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.8,
74
+ "at_end_success_rate": 0.8,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.8
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.92,
81
+ "current_chain_depth_progress_score": 0.92,
82
+ "latched_chain_depth_progress_score": 0.92,
83
+ "avg_current_chain_depth": 2.76,
84
+ "avg_latched_chain_depth": 2.76
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.0416082728899971,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2083.05,
91
+ "avg_policy_steps_to_success": 694.7,
92
+ "avg_policy_queries_to_success": 694.7
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 5.182152666217547e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_3": 3
98
+ },
99
+ "robot_constraint_violation_rate": 0.04,
100
+ "finger_joint_limit_saturation_rate": 0.0,
101
+ "joint_limit_active_step_rate": 0.0,
102
+ "finger_joint_limit_saturation_step_rate": 0.0,
103
+ "max_joint_limit_excess_rad": 0.005119434092193842,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {},
106
+ "joint_limit_hard_counts": {
107
+ "multi_ring_q2": 1
108
+ },
109
+ "joint_limit_hard_episode_events": 1,
110
+ "finger_joint_limit_saturation_counts": {},
111
+ "joint_limit_max_excess_by_joint": {
112
+ "multi_pinky_q1": 0.001346059376373887,
113
+ "multi_pinky_q2": 0.0036378963850438595,
114
+ "multi_ring_q1": 0.0016813931288197637,
115
+ "multi_ring_q2": 0.005119434092193842,
116
+ "pinky_q1": 0.0004963836981914937,
117
+ "pinky_q2": 0.0029781206976622343,
118
+ "ring_q1": 0.0010066964896395802,
119
+ "ring_q2": 0.004073512274771929
120
+ },
121
+ "coverage": {
122
+ "hard_limit_source": {
123
+ "available": 25,
124
+ "unavailable": 0,
125
+ "unknown": 0
126
+ },
127
+ "joint_role_manifest": {
128
+ "available": 25,
129
+ "unavailable": 0,
130
+ "unknown": 0
131
+ }
132
+ },
133
+ "hard_limit_sources": {
134
+ "runtime.joint_pos_limits": 25
135
+ },
136
+ "soft_limit_sources": {
137
+ "runtime.soft_joint_pos_limits": 25
138
+ },
139
+ "unknown_runtime_joint_names": [],
140
+ "baseline_checks": {
141
+ "post_home": {
142
+ "checked": 25,
143
+ "hard_violation_episodes": 0
144
+ },
145
+ "post_reset": {
146
+ "checked": 25,
147
+ "hard_violation_episodes": 0
148
+ }
149
+ }
150
+ }
151
+ },
152
+ "robot_motion": {
153
+ "joint_acceleration_rms": 7.746263423255011,
154
+ "joint_effort_rms": 84.34382686600972,
155
+ "joint_jerk_rms": 786.9518707232774,
156
+ "joint_velocity_rms": 0.474268390267282
157
+ },
158
+ "grasp": {
159
+ "kinematic_grasp_stability_index": 0.2877930899587484,
160
+ "mean_kinematic_grasp_stability": 0.07086260407346791,
161
+ "per_object_best_kinematic_grasp_stability": {
162
+ "obj_194_wineglass_1": 0.1195839269119183,
163
+ "obj_194_wineglass_2": 0.04867322153981351,
164
+ "obj_194_wineglass_3": 0.2196457444782759
165
+ },
166
+ "grasp_gsi_diagnostics": {
167
+ "mean_anti_slip_score": 0.9641718496358521,
168
+ "mean_duration_score": 0.06695133120179232,
169
+ "mean_hold_score": 0.06976982138075452,
170
+ "mean_kinematic_stability_score": 0.021697034745075137,
171
+ "mean_legacy_gsi": 0.26200734524412783,
172
+ "mean_lift_score": 0.21421170971676579,
173
+ "mean_motion_stability_score": 0.07224631039572163
174
+ }
175
+ },
176
+ "tool": {
177
+ "tool_selection_accuracy": null,
178
+ "tool_switch_success_rate": null,
179
+ "tool_switch_total": 0,
180
+ "tool_switch_successful": 0
181
+ },
182
+ "runtime": {
183
+ "avg_steps": 2315.64,
184
+ "error_count": 0
185
+ }
186
+ }
187
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Full/summary.json ADDED
@@ -0,0 +1,188 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 200300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 40,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.8,
50
+ "success_rate_ci": {
51
+ "mean": 0.8,
52
+ "ci_low": 0.68,
53
+ "ci_high": 0.9
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.9266666666666666
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 35.00166666666596
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.78,
64
+ "hard_violation_rate": 0.08,
65
+ "drop_rate": 0.08,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 3.434537711224069e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.82,
74
+ "at_end_success_rate": 0.8,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.8
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.9266666666666666,
81
+ "current_chain_depth_progress_score": 0.9266666666666666,
82
+ "latched_chain_depth_progress_score": 0.9266666666666666,
83
+ "avg_current_chain_depth": 2.78,
84
+ "avg_latched_chain_depth": 2.78
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.0333852733956834,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2100.1,
91
+ "avg_policy_steps_to_success": 700.35,
92
+ "avg_policy_queries_to_success": 700.35
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 3.434537711224069e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_1": 1,
98
+ "drop_to_table_obj_194_wineglass_3": 3
99
+ },
100
+ "robot_constraint_violation_rate": 0.02,
101
+ "finger_joint_limit_saturation_rate": 0.0,
102
+ "joint_limit_active_step_rate": 0.0,
103
+ "finger_joint_limit_saturation_step_rate": 0.0,
104
+ "max_joint_limit_excess_rad": 0.005119434092193842,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {},
107
+ "joint_limit_hard_counts": {
108
+ "multi_ring_q2": 1
109
+ },
110
+ "joint_limit_hard_episode_events": 1,
111
+ "finger_joint_limit_saturation_counts": {},
112
+ "joint_limit_max_excess_by_joint": {
113
+ "multi_pinky_q1": 0.0017718856688588858,
114
+ "multi_pinky_q2": 0.0036378963850438595,
115
+ "multi_ring_q1": 0.0017534798244014382,
116
+ "multi_ring_q2": 0.005119434092193842,
117
+ "pinky_q1": 0.0005137042608112097,
118
+ "pinky_q2": 0.0029781206976622343,
119
+ "ring_q1": 0.0010066964896395802,
120
+ "ring_q2": 0.004073512274771929
121
+ },
122
+ "coverage": {
123
+ "hard_limit_source": {
124
+ "available": 50,
125
+ "unavailable": 0,
126
+ "unknown": 0
127
+ },
128
+ "joint_role_manifest": {
129
+ "available": 50,
130
+ "unavailable": 0,
131
+ "unknown": 0
132
+ }
133
+ },
134
+ "hard_limit_sources": {
135
+ "runtime.joint_pos_limits": 50
136
+ },
137
+ "soft_limit_sources": {
138
+ "runtime.soft_joint_pos_limits": 50
139
+ },
140
+ "unknown_runtime_joint_names": [],
141
+ "baseline_checks": {
142
+ "post_home": {
143
+ "checked": 50,
144
+ "hard_violation_episodes": 0
145
+ },
146
+ "post_reset": {
147
+ "checked": 50,
148
+ "hard_violation_episodes": 0
149
+ }
150
+ }
151
+ }
152
+ },
153
+ "robot_motion": {
154
+ "joint_acceleration_rms": 7.727971984322354,
155
+ "joint_effort_rms": 83.75094540848646,
156
+ "joint_jerk_rms": 785.0191286916842,
157
+ "joint_velocity_rms": 0.4760206285831818
158
+ },
159
+ "grasp": {
160
+ "kinematic_grasp_stability_index": 0.3426681901476394,
161
+ "mean_kinematic_grasp_stability": 0.1072025025967074,
162
+ "per_object_best_kinematic_grasp_stability": {
163
+ "obj_194_wineglass_1": 0.1493049048897465,
164
+ "obj_194_wineglass_2": 0.05830824329390662,
165
+ "obj_194_wineglass_3": 0.2562279670705605
166
+ },
167
+ "grasp_gsi_diagnostics": {
168
+ "mean_anti_slip_score": 0.9560947620742114,
169
+ "mean_duration_score": 0.11139155575636442,
170
+ "mean_hold_score": 0.0671578105627825,
171
+ "mean_kinematic_stability_score": 0.029917477291034264,
172
+ "mean_legacy_gsi": 0.26450782890856545,
173
+ "mean_lift_score": 0.18947479329988587,
174
+ "mean_motion_stability_score": 0.0662106818607422
175
+ }
176
+ },
177
+ "tool": {
178
+ "tool_selection_accuracy": null,
179
+ "tool_switch_success_rate": null,
180
+ "tool_switch_total": 0,
181
+ "tool_switch_successful": 0
182
+ },
183
+ "runtime": {
184
+ "avg_steps": 2329.28,
185
+ "error_count": 0
186
+ }
187
+ }
188
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/native_shards/000/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/native_shards/000/summary.json ADDED
@@ -0,0 +1,187 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 100300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 17,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.68,
50
+ "success_rate_ci": {
51
+ "mean": 0.68,
52
+ "ci_low": 0.48,
53
+ "ci_high": 0.84
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.9333333333333333
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 33.82745098039152
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.64,
64
+ "hard_violation_rate": 0.2,
65
+ "drop_rate": 0.2,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 8.268289456277285e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.68,
74
+ "at_end_success_rate": 0.68,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.68
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.92,
81
+ "current_chain_depth_progress_score": 0.92,
82
+ "latched_chain_depth_progress_score": 0.9333333333333333,
83
+ "avg_current_chain_depth": 2.76,
84
+ "avg_latched_chain_depth": 2.8
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.0677567854091485,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2029.6470588235295,
91
+ "avg_policy_steps_to_success": 676.8823529411765,
92
+ "avg_policy_queries_to_success": 676.8823529411765
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 8.268289456277285e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_3": 5
98
+ },
99
+ "robot_constraint_violation_rate": 0.04,
100
+ "finger_joint_limit_saturation_rate": 0.0,
101
+ "joint_limit_active_step_rate": 0.0,
102
+ "finger_joint_limit_saturation_step_rate": 0.0,
103
+ "max_joint_limit_excess_rad": 0.005090742837637663,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {},
106
+ "joint_limit_hard_counts": {
107
+ "multi_ring_q2": 1
108
+ },
109
+ "joint_limit_hard_episode_events": 1,
110
+ "finger_joint_limit_saturation_counts": {},
111
+ "joint_limit_max_excess_by_joint": {
112
+ "multi_pinky_q1": 0.0008058245293796062,
113
+ "multi_pinky_q2": 0.003547410713508725,
114
+ "multi_ring_q1": 0.0007777615683153272,
115
+ "multi_ring_q2": 0.005090742837637663,
116
+ "pinky_q1": 0.0007292705122381449,
117
+ "pinky_q2": 0.002905200934037566,
118
+ "ring_q1": 0.0013935439055785537,
119
+ "ring_q2": 0.003969747573137283
120
+ },
121
+ "coverage": {
122
+ "hard_limit_source": {
123
+ "available": 25,
124
+ "unavailable": 0,
125
+ "unknown": 0
126
+ },
127
+ "joint_role_manifest": {
128
+ "available": 25,
129
+ "unavailable": 0,
130
+ "unknown": 0
131
+ }
132
+ },
133
+ "hard_limit_sources": {
134
+ "runtime.joint_pos_limits": 25
135
+ },
136
+ "soft_limit_sources": {
137
+ "runtime.soft_joint_pos_limits": 25
138
+ },
139
+ "unknown_runtime_joint_names": [],
140
+ "baseline_checks": {
141
+ "post_home": {
142
+ "checked": 25,
143
+ "hard_violation_episodes": 0
144
+ },
145
+ "post_reset": {
146
+ "checked": 25,
147
+ "hard_violation_episodes": 0
148
+ }
149
+ }
150
+ }
151
+ },
152
+ "robot_motion": {
153
+ "joint_acceleration_rms": 7.824105726348113,
154
+ "joint_effort_rms": 83.32838694331447,
155
+ "joint_jerk_rms": 798.3230561046158,
156
+ "joint_velocity_rms": 0.47264383774966096
157
+ },
158
+ "grasp": {
159
+ "kinematic_grasp_stability_index": 0.45050648410893124,
160
+ "mean_kinematic_grasp_stability": 0.19264972762528743,
161
+ "per_object_best_kinematic_grasp_stability": {
162
+ "obj_194_wineglass_1": 0.10803441774524361,
163
+ "obj_194_wineglass_2": 0.1455324753514822,
164
+ "obj_194_wineglass_3": 0.2900873234836258
165
+ },
166
+ "grasp_gsi_diagnostics": {
167
+ "mean_anti_slip_score": 0.9742237564295296,
168
+ "mean_duration_score": 0.10067973736175517,
169
+ "mean_hold_score": 0.07365492932463062,
170
+ "mean_kinematic_stability_score": 0.05344522006671704,
171
+ "mean_legacy_gsi": 0.26970598785980165,
172
+ "mean_lift_score": 0.18222940561762083,
173
+ "mean_motion_stability_score": 0.07618766985811795
174
+ }
175
+ },
176
+ "tool": {
177
+ "tool_selection_accuracy": null,
178
+ "tool_switch_success_rate": null,
179
+ "tool_switch_total": 0,
180
+ "tool_switch_successful": 0
181
+ },
182
+ "runtime": {
183
+ "avg_steps": 2418.88,
184
+ "error_count": 0
185
+ }
186
+ }
187
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/native_shards/001/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/native_shards/001/summary.json ADDED
@@ -0,0 +1,193 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 100300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 20,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.8,
50
+ "success_rate_ci": {
51
+ "mean": 0.8,
52
+ "ci_low": 0.64,
53
+ "ci_high": 0.96
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.9066666666666666
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 34.503333333332655
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.8,
64
+ "hard_violation_rate": 0.08,
65
+ "drop_rate": 0.08,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 3.470173855710171e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.8,
74
+ "at_end_success_rate": 0.8,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.8
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.8933333333333333,
81
+ "current_chain_depth_progress_score": 0.8933333333333333,
82
+ "latched_chain_depth_progress_score": 0.9066666666666666,
83
+ "avg_current_chain_depth": 2.68,
84
+ "avg_latched_chain_depth": 2.72
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.0478647409368915,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2070.2,
91
+ "avg_policy_steps_to_success": 690.45,
92
+ "avg_policy_queries_to_success": 690.45
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 3.470173855710171e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_1": 1,
98
+ "drop_to_table_obj_194_wineglass_3": 1
99
+ },
100
+ "robot_constraint_violation_rate": 0.0,
101
+ "finger_joint_limit_saturation_rate": 0.0,
102
+ "joint_limit_active_step_rate": 0.0,
103
+ "finger_joint_limit_saturation_step_rate": 0.0,
104
+ "max_joint_limit_excess_rad": 0.00478564016520977,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {},
107
+ "joint_limit_hard_counts": {},
108
+ "joint_limit_hard_episode_events": 0,
109
+ "finger_joint_limit_saturation_counts": {},
110
+ "joint_limit_max_excess_by_joint": {
111
+ "index_q1": 5.268533387337015e-10,
112
+ "index_q2": 7.266733326183328e-10,
113
+ "multi_middle_q1": 5.784106527073618e-10,
114
+ "multi_pinky_q1": 0.0012199339689686894,
115
+ "multi_pinky_q2": 0.0033348114229738712,
116
+ "multi_ring_q1": 0.0014034457271918654,
117
+ "multi_ring_q2": 0.00478564016520977,
118
+ "multi_thumb_q1": 1.1553316081602816e-09,
119
+ "multi_thumb_q2": 1.3309879587364293e-10,
120
+ "pinky_q1": 0.000862616638187319,
121
+ "pinky_q2": 0.0029347382951527834,
122
+ "ring_q1": 0.0010783164761960506,
123
+ "ring_q2": 0.004085349850356579,
124
+ "thumb_q1": 7.678177538217312e-10,
125
+ "thumb_q2": 3.740576459687617e-10
126
+ },
127
+ "coverage": {
128
+ "hard_limit_source": {
129
+ "available": 25,
130
+ "unavailable": 0,
131
+ "unknown": 0
132
+ },
133
+ "joint_role_manifest": {
134
+ "available": 25,
135
+ "unavailable": 0,
136
+ "unknown": 0
137
+ }
138
+ },
139
+ "hard_limit_sources": {
140
+ "runtime.joint_pos_limits": 25
141
+ },
142
+ "soft_limit_sources": {
143
+ "runtime.soft_joint_pos_limits": 25
144
+ },
145
+ "unknown_runtime_joint_names": [],
146
+ "baseline_checks": {
147
+ "post_home": {
148
+ "checked": 25,
149
+ "hard_violation_episodes": 0
150
+ },
151
+ "post_reset": {
152
+ "checked": 25,
153
+ "hard_violation_episodes": 0
154
+ }
155
+ }
156
+ }
157
+ },
158
+ "robot_motion": {
159
+ "joint_acceleration_rms": 7.392360583057148,
160
+ "joint_effort_rms": 83.91258503673106,
161
+ "joint_jerk_rms": 750.7692490400519,
162
+ "joint_velocity_rms": 0.4669447310563978
163
+ },
164
+ "grasp": {
165
+ "kinematic_grasp_stability_index": 0.3008126697655456,
166
+ "mean_kinematic_grasp_stability": 0.11465472029895561,
167
+ "per_object_best_kinematic_grasp_stability": {
168
+ "obj_194_wineglass_1": 0.14562626919902016,
169
+ "obj_194_wineglass_2": 0.123354594667729,
170
+ "obj_194_wineglass_3": 0.17268516460723277
171
+ },
172
+ "grasp_gsi_diagnostics": {
173
+ "mean_anti_slip_score": 0.9758624191538832,
174
+ "mean_duration_score": 0.103686965341827,
175
+ "mean_hold_score": 0.060421397147296364,
176
+ "mean_kinematic_stability_score": 0.039395662544174,
177
+ "mean_legacy_gsi": 0.26273365676589305,
178
+ "mean_lift_score": 0.17593268013006444,
179
+ "mean_motion_stability_score": 0.06052481929859962
180
+ }
181
+ },
182
+ "tool": {
183
+ "tool_selection_accuracy": null,
184
+ "tool_switch_success_rate": null,
185
+ "tool_switch_total": 0,
186
+ "tool_switch_successful": 0
187
+ },
188
+ "runtime": {
189
+ "avg_steps": 2305.36,
190
+ "error_count": 0
191
+ }
192
+ }
193
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/Inv/summary.json ADDED
@@ -0,0 +1,195 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 100300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 37,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.74,
50
+ "success_rate_ci": {
51
+ "mean": 0.74,
52
+ "ci_low": 0.62,
53
+ "ci_high": 0.86
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.92
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 34.19279279279213
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.72,
64
+ "hard_violation_rate": 0.14,
65
+ "drop_rate": 0.14,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 5.926879244068887e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.74,
74
+ "at_end_success_rate": 0.74,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.74
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.9066666666666666,
81
+ "current_chain_depth_progress_score": 0.9066666666666666,
82
+ "latched_chain_depth_progress_score": 0.92,
83
+ "avg_current_chain_depth": 2.72,
84
+ "avg_latched_chain_depth": 2.76
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.0570043289376583,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2051.5675675675675,
91
+ "avg_policy_steps_to_success": 684.2162162162163,
92
+ "avg_policy_queries_to_success": 684.2162162162163
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 5.926879244068887e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_1": 1,
98
+ "drop_to_table_obj_194_wineglass_3": 6
99
+ },
100
+ "robot_constraint_violation_rate": 0.02,
101
+ "finger_joint_limit_saturation_rate": 0.0,
102
+ "joint_limit_active_step_rate": 0.0,
103
+ "finger_joint_limit_saturation_step_rate": 0.0,
104
+ "max_joint_limit_excess_rad": 0.005090742837637663,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {},
107
+ "joint_limit_hard_counts": {
108
+ "multi_ring_q2": 1
109
+ },
110
+ "joint_limit_hard_episode_events": 1,
111
+ "finger_joint_limit_saturation_counts": {},
112
+ "joint_limit_max_excess_by_joint": {
113
+ "index_q1": 5.268533387337015e-10,
114
+ "index_q2": 7.266733326183328e-10,
115
+ "multi_middle_q1": 5.784106527073618e-10,
116
+ "multi_pinky_q1": 0.0012199339689686894,
117
+ "multi_pinky_q2": 0.003547410713508725,
118
+ "multi_ring_q1": 0.0014034457271918654,
119
+ "multi_ring_q2": 0.005090742837637663,
120
+ "multi_thumb_q1": 1.1553316081602816e-09,
121
+ "multi_thumb_q2": 1.3309879587364293e-10,
122
+ "pinky_q1": 0.000862616638187319,
123
+ "pinky_q2": 0.0029347382951527834,
124
+ "ring_q1": 0.0013935439055785537,
125
+ "ring_q2": 0.004085349850356579,
126
+ "thumb_q1": 7.678177538217312e-10,
127
+ "thumb_q2": 3.740576459687617e-10
128
+ },
129
+ "coverage": {
130
+ "hard_limit_source": {
131
+ "available": 50,
132
+ "unavailable": 0,
133
+ "unknown": 0
134
+ },
135
+ "joint_role_manifest": {
136
+ "available": 50,
137
+ "unavailable": 0,
138
+ "unknown": 0
139
+ }
140
+ },
141
+ "hard_limit_sources": {
142
+ "runtime.joint_pos_limits": 50
143
+ },
144
+ "soft_limit_sources": {
145
+ "runtime.soft_joint_pos_limits": 50
146
+ },
147
+ "unknown_runtime_joint_names": [],
148
+ "baseline_checks": {
149
+ "post_home": {
150
+ "checked": 50,
151
+ "hard_violation_episodes": 0
152
+ },
153
+ "post_reset": {
154
+ "checked": 50,
155
+ "hard_violation_episodes": 0
156
+ }
157
+ }
158
+ }
159
+ },
160
+ "robot_motion": {
161
+ "joint_acceleration_rms": 7.6082331547026305,
162
+ "joint_effort_rms": 83.62048599002277,
163
+ "joint_jerk_rms": 774.5461525723339,
164
+ "joint_velocity_rms": 0.46979428440302934
165
+ },
166
+ "grasp": {
167
+ "kinematic_grasp_stability_index": 0.3756595769372384,
168
+ "mean_kinematic_grasp_stability": 0.1544480913838188,
169
+ "per_object_best_kinematic_grasp_stability": {
170
+ "obj_194_wineglass_1": 0.12683034347213187,
171
+ "obj_194_wineglass_2": 0.1344435350096056,
172
+ "obj_194_wineglass_3": 0.2313862440454293
173
+ },
174
+ "grasp_gsi_diagnostics": {
175
+ "mean_anti_slip_score": 0.9750430877917063,
176
+ "mean_duration_score": 0.10218335135179109,
177
+ "mean_hold_score": 0.0670381632359635,
178
+ "mean_kinematic_stability_score": 0.04642044130544552,
179
+ "mean_legacy_gsi": 0.26621982231284735,
180
+ "mean_lift_score": 0.17908104287384263,
181
+ "mean_motion_stability_score": 0.06835624457835879
182
+ }
183
+ },
184
+ "tool": {
185
+ "tool_selection_accuracy": null,
186
+ "tool_switch_success_rate": null,
187
+ "tool_switch_total": 0,
188
+ "tool_switch_successful": 0
189
+ },
190
+ "runtime": {
191
+ "avg_steps": 2362.12,
192
+ "error_count": 0
193
+ }
194
+ }
195
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/native_shards/000/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/native_shards/000/summary.json ADDED
@@ -0,0 +1,185 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 100300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 20,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.8,
50
+ "success_rate_ci": {
51
+ "mean": 0.8,
52
+ "ci_low": 0.64,
53
+ "ci_high": 0.96
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.96
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 33.97666666666602
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.76,
64
+ "hard_violation_rate": 0.04,
65
+ "drop_rate": 0.04,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 1.754324409669836e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.88,
74
+ "at_end_success_rate": 0.8,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.8
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.96,
81
+ "current_chain_depth_progress_score": 0.96,
82
+ "latched_chain_depth_progress_score": 0.96,
83
+ "avg_current_chain_depth": 2.88,
84
+ "avg_latched_chain_depth": 2.88
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.0627627065364247,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2038.6,
91
+ "avg_policy_steps_to_success": 679.8,
92
+ "avg_policy_queries_to_success": 679.8
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 1.754324409669836e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_3": 1
98
+ },
99
+ "robot_constraint_violation_rate": 0.0,
100
+ "finger_joint_limit_saturation_rate": 0.0,
101
+ "joint_limit_active_step_rate": 0.0,
102
+ "finger_joint_limit_saturation_step_rate": 0.0,
103
+ "max_joint_limit_excess_rad": 0.00432391744107008,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {},
106
+ "joint_limit_hard_counts": {},
107
+ "joint_limit_hard_episode_events": 0,
108
+ "finger_joint_limit_saturation_counts": {},
109
+ "joint_limit_max_excess_by_joint": {
110
+ "multi_pinky_q1": 0.001048040809109807,
111
+ "multi_pinky_q2": 0.003485720371827483,
112
+ "multi_ring_q1": 0.0008981552091427147,
113
+ "multi_ring_q2": 0.00432391744107008,
114
+ "pinky_q1": 0.0005132189253345132,
115
+ "pinky_q2": 0.0026407288387417793,
116
+ "ring_q1": 0.0009725192212499678,
117
+ "ring_q2": 0.003572287503629923
118
+ },
119
+ "coverage": {
120
+ "hard_limit_source": {
121
+ "available": 25,
122
+ "unavailable": 0,
123
+ "unknown": 0
124
+ },
125
+ "joint_role_manifest": {
126
+ "available": 25,
127
+ "unavailable": 0,
128
+ "unknown": 0
129
+ }
130
+ },
131
+ "hard_limit_sources": {
132
+ "runtime.joint_pos_limits": 25
133
+ },
134
+ "soft_limit_sources": {
135
+ "runtime.soft_joint_pos_limits": 25
136
+ },
137
+ "unknown_runtime_joint_names": [],
138
+ "baseline_checks": {
139
+ "post_home": {
140
+ "checked": 25,
141
+ "hard_violation_episodes": 0
142
+ },
143
+ "post_reset": {
144
+ "checked": 25,
145
+ "hard_violation_episodes": 0
146
+ }
147
+ }
148
+ }
149
+ },
150
+ "robot_motion": {
151
+ "joint_acceleration_rms": 7.998968244206217,
152
+ "joint_effort_rms": 84.47687357898384,
153
+ "joint_jerk_rms": 816.328187703419,
154
+ "joint_velocity_rms": 0.4831393878828024
155
+ },
156
+ "grasp": {
157
+ "kinematic_grasp_stability_index": 0.31648032181874347,
158
+ "mean_kinematic_grasp_stability": 0.0992218953667017,
159
+ "per_object_best_kinematic_grasp_stability": {
160
+ "obj_194_wineglass_1": 0.09297950386629468,
161
+ "obj_194_wineglass_2": 0.12270080250092118,
162
+ "obj_194_wineglass_3": 0.23918443705777487
163
+ },
164
+ "grasp_gsi_diagnostics": {
165
+ "mean_anti_slip_score": 0.9505174459239124,
166
+ "mean_duration_score": 0.11671165098877061,
167
+ "mean_hold_score": 0.05768130058675762,
168
+ "mean_kinematic_stability_score": 0.0287556155031202,
169
+ "mean_legacy_gsi": 0.25858382573663574,
170
+ "mean_lift_score": 0.18255371234841322,
171
+ "mean_motion_stability_score": 0.05499870149548098
172
+ }
173
+ },
174
+ "tool": {
175
+ "tool_selection_accuracy": null,
176
+ "tool_switch_success_rate": null,
177
+ "tool_switch_total": 0,
178
+ "tool_switch_successful": 0
179
+ },
180
+ "runtime": {
181
+ "avg_steps": 2280.08,
182
+ "error_count": 0
183
+ }
184
+ }
185
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/native_shards/001/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/native_shards/001/summary.json ADDED
@@ -0,0 +1,189 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 100300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 19,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.76,
50
+ "success_rate_ci": {
51
+ "mean": 0.76,
52
+ "ci_low": 0.6,
53
+ "ci_high": 0.92
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.9333333333333333
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 34.601754385964234
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.68,
64
+ "hard_violation_rate": 0.16,
65
+ "drop_rate": 0.16,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 8.485794779539052e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.76,
74
+ "at_end_success_rate": 0.76,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.76
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.9333333333333333,
81
+ "current_chain_depth_progress_score": 0.9333333333333333,
82
+ "latched_chain_depth_progress_score": 0.9333333333333333,
83
+ "avg_current_chain_depth": 2.8,
84
+ "avg_latched_chain_depth": 2.8
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.045445860826074,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2076.1052631578946,
91
+ "avg_policy_steps_to_success": 692.421052631579,
92
+ "avg_policy_queries_to_success": 692.421052631579
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 8.485794779539052e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_1": 1,
98
+ "drop_to_table_obj_194_wineglass_2": 1,
99
+ "drop_to_table_obj_194_wineglass_3": 3
100
+ },
101
+ "robot_constraint_violation_rate": 0.04,
102
+ "finger_joint_limit_saturation_rate": 0.0,
103
+ "joint_limit_active_step_rate": 0.0,
104
+ "finger_joint_limit_saturation_step_rate": 0.0,
105
+ "max_joint_limit_excess_rad": 0.005265207029879093,
106
+ "robot_constraint_diagnostics": {
107
+ "joint_limit_soft_counts": {},
108
+ "joint_limit_hard_counts": {
109
+ "multi_ring_q2": 1
110
+ },
111
+ "joint_limit_hard_episode_events": 1,
112
+ "finger_joint_limit_saturation_counts": {},
113
+ "joint_limit_max_excess_by_joint": {
114
+ "multi_pinky_q1": 0.002401346806436777,
115
+ "multi_pinky_q2": 0.0037128350231796503,
116
+ "multi_ring_q1": 0.0024835264775902033,
117
+ "multi_ring_q2": 0.005265207029879093,
118
+ "pinky_q1": 0.0005623262841254473,
119
+ "pinky_q2": 0.002649222034960985,
120
+ "ring_q1": 0.0012508858926594257,
121
+ "ring_q2": 0.0038569460157305002
122
+ },
123
+ "coverage": {
124
+ "hard_limit_source": {
125
+ "available": 25,
126
+ "unavailable": 0,
127
+ "unknown": 0
128
+ },
129
+ "joint_role_manifest": {
130
+ "available": 25,
131
+ "unavailable": 0,
132
+ "unknown": 0
133
+ }
134
+ },
135
+ "hard_limit_sources": {
136
+ "runtime.joint_pos_limits": 25
137
+ },
138
+ "soft_limit_sources": {
139
+ "runtime.soft_joint_pos_limits": 25
140
+ },
141
+ "unknown_runtime_joint_names": [],
142
+ "baseline_checks": {
143
+ "post_home": {
144
+ "checked": 25,
145
+ "hard_violation_episodes": 0
146
+ },
147
+ "post_reset": {
148
+ "checked": 25,
149
+ "hard_violation_episodes": 0
150
+ }
151
+ }
152
+ }
153
+ },
154
+ "robot_motion": {
155
+ "joint_acceleration_rms": 7.614557492631298,
156
+ "joint_effort_rms": 84.16191524319811,
157
+ "joint_jerk_rms": 773.2253043945509,
158
+ "joint_velocity_rms": 0.47879316595860594
159
+ },
160
+ "grasp": {
161
+ "kinematic_grasp_stability_index": 0.3882300395757969,
162
+ "mean_kinematic_grasp_stability": 0.156951698917567,
163
+ "per_object_best_kinematic_grasp_stability": {
164
+ "obj_194_wineglass_1": 0.15368154192223937,
165
+ "obj_194_wineglass_2": 0.15994127256825955,
166
+ "obj_194_wineglass_3": 0.2110354460257855
167
+ },
168
+ "grasp_gsi_diagnostics": {
169
+ "mean_anti_slip_score": 0.953119518057465,
170
+ "mean_duration_score": 0.13449544714099657,
171
+ "mean_hold_score": 0.0743492417462408,
172
+ "mean_kinematic_stability_score": 0.05015120388364321,
173
+ "mean_legacy_gsi": 0.2750473048103452,
174
+ "mean_lift_score": 0.20781126009348674,
175
+ "mean_motion_stability_score": 0.07722057471455156
176
+ }
177
+ },
178
+ "tool": {
179
+ "tool_selection_accuracy": null,
180
+ "tool_switch_success_rate": null,
181
+ "tool_switch_total": 0,
182
+ "tool_switch_successful": 0
183
+ },
184
+ "runtime": {
185
+ "avg_steps": 2356.88,
186
+ "error_count": 0
187
+ }
188
+ }
189
+ }
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/03_wine_glass_plate_balance/None/summary.json ADDED
@@ -0,0 +1,189 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA wine_fast-e9124b6af8",
5
+ "base_seed": 100300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 39,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 3246,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.78,
50
+ "success_rate_ci": {
51
+ "mean": 0.78,
52
+ "ci_low": 0.66,
53
+ "ci_high": 0.88
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.9466666666666667
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 34.28119658119592
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.72,
64
+ "hard_violation_rate": 0.1,
65
+ "drop_rate": 0.1,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 5.175804837652255e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.82,
74
+ "at_end_success_rate": 0.78,
75
+ "task_family_success_rate": {
76
+ "table_setting": 0.78
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.9466666666666667,
81
+ "current_chain_depth_progress_score": 0.9466666666666667,
82
+ "latched_chain_depth_progress_score": 0.9466666666666667,
83
+ "avg_current_chain_depth": 2.84,
84
+ "avg_latched_chain_depth": 2.84
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 1.0543262945236898,
88
+ "expert_time_s": 36.06666666666666,
89
+ "expert_time_step": 2164,
90
+ "avg_steps_to_success": 2056.871794871795,
91
+ "avg_policy_steps_to_success": 685.9487179487179,
92
+ "avg_policy_queries_to_success": 685.9487179487179
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 5.175804837652255e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_194_wineglass_1": 1,
98
+ "drop_to_table_obj_194_wineglass_2": 1,
99
+ "drop_to_table_obj_194_wineglass_3": 4
100
+ },
101
+ "robot_constraint_violation_rate": 0.02,
102
+ "finger_joint_limit_saturation_rate": 0.0,
103
+ "joint_limit_active_step_rate": 0.0,
104
+ "finger_joint_limit_saturation_step_rate": 0.0,
105
+ "max_joint_limit_excess_rad": 0.005265207029879093,
106
+ "robot_constraint_diagnostics": {
107
+ "joint_limit_soft_counts": {},
108
+ "joint_limit_hard_counts": {
109
+ "multi_ring_q2": 1
110
+ },
111
+ "joint_limit_hard_episode_events": 1,
112
+ "finger_joint_limit_saturation_counts": {},
113
+ "joint_limit_max_excess_by_joint": {
114
+ "multi_pinky_q1": 0.002401346806436777,
115
+ "multi_pinky_q2": 0.0037128350231796503,
116
+ "multi_ring_q1": 0.0024835264775902033,
117
+ "multi_ring_q2": 0.005265207029879093,
118
+ "pinky_q1": 0.0005623262841254473,
119
+ "pinky_q2": 0.002649222034960985,
120
+ "ring_q1": 0.0012508858926594257,
121
+ "ring_q2": 0.0038569460157305002
122
+ },
123
+ "coverage": {
124
+ "hard_limit_source": {
125
+ "available": 50,
126
+ "unavailable": 0,
127
+ "unknown": 0
128
+ },
129
+ "joint_role_manifest": {
130
+ "available": 50,
131
+ "unavailable": 0,
132
+ "unknown": 0
133
+ }
134
+ },
135
+ "hard_limit_sources": {
136
+ "runtime.joint_pos_limits": 50
137
+ },
138
+ "soft_limit_sources": {
139
+ "runtime.soft_joint_pos_limits": 50
140
+ },
141
+ "unknown_runtime_joint_names": [],
142
+ "baseline_checks": {
143
+ "post_home": {
144
+ "checked": 50,
145
+ "hard_violation_episodes": 0
146
+ },
147
+ "post_reset": {
148
+ "checked": 50,
149
+ "hard_violation_episodes": 0
150
+ }
151
+ }
152
+ }
153
+ },
154
+ "robot_motion": {
155
+ "joint_acceleration_rms": 7.806762868418757,
156
+ "joint_effort_rms": 84.31939441109098,
157
+ "joint_jerk_rms": 794.7767460489849,
158
+ "joint_velocity_rms": 0.4809662769207042
159
+ },
160
+ "grasp": {
161
+ "kinematic_grasp_stability_index": 0.3523551806972702,
162
+ "mean_kinematic_grasp_stability": 0.12808679714213433,
163
+ "per_object_best_kinematic_grasp_stability": {
164
+ "obj_194_wineglass_1": 0.12333052289426702,
165
+ "obj_194_wineglass_2": 0.14132103753459035,
166
+ "obj_194_wineglass_3": 0.22510994154178018
167
+ },
168
+ "grasp_gsi_diagnostics": {
169
+ "mean_anti_slip_score": 0.9518184819906886,
170
+ "mean_duration_score": 0.12560354906488358,
171
+ "mean_hold_score": 0.0660152711664992,
172
+ "mean_kinematic_stability_score": 0.03945340969338171,
173
+ "mean_legacy_gsi": 0.26681556527349043,
174
+ "mean_lift_score": 0.19518248622095,
175
+ "mean_motion_stability_score": 0.06610963810501627
176
+ }
177
+ },
178
+ "tool": {
179
+ "tool_selection_accuracy": null,
180
+ "tool_switch_success_rate": null,
181
+ "tool_switch_total": 0,
182
+ "tool_switch_successful": 0
183
+ },
184
+ "runtime": {
185
+ "avg_steps": 2318.48,
186
+ "error_count": 0
187
+ }
188
+ }
189
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/native_shards/000/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/native_shards/000/summary.json ADDED
@@ -0,0 +1,222 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 100600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 16,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.64,
50
+ "success_rate_ci": {
51
+ "mean": 0.64,
52
+ "ci_low": 0.44,
53
+ "ci_high": 0.84
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.82
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.157291666666236
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.64,
64
+ "hard_violation_rate": 0.08,
65
+ "drop_rate": 0.08,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 4.3261015336029936e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.64,
74
+ "at_end_success_rate": 0.64,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.64
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.81,
81
+ "current_chain_depth_progress_score": 0.8533333333333333,
82
+ "latched_chain_depth_progress_score": 0.8666666666666667,
83
+ "avg_current_chain_depth": 2.56,
84
+ "avg_latched_chain_depth": 2.6
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7077753960255263,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1809.4375,
91
+ "avg_policy_steps_to_success": 603.5,
92
+ "avg_policy_queries_to_success": 603.5
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 4.3261015336029936e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 1,
98
+ "drop_to_table_obj_013_apple_4": 1
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.28,
102
+ "joint_limit_active_step_rate": 0.0920378101274037,
103
+ "finger_joint_limit_saturation_step_rate": 0.005169691332655577,
104
+ "max_joint_limit_excess_rad": 0.6389230489730835,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "left_index_1_joint": 18,
108
+ "left_middle_1_joint": 6,
109
+ "left_ring_1_joint": 36,
110
+ "left_thumb_2_joint": 107,
111
+ "right_little_1_joint": 886,
112
+ "right_middle_1_joint": 32,
113
+ "right_thumb_2_joint": 3356
114
+ },
115
+ "joint_limit_hard_counts": {
116
+ "left_index_1_joint": 1,
117
+ "left_little_1_joint": 5,
118
+ "left_middle_1_joint": 1,
119
+ "left_ring_1_joint": 6,
120
+ "left_thumb_2_joint": 34,
121
+ "right_little_1_joint": 43,
122
+ "right_middle_1_joint": 4,
123
+ "right_thumb_2_joint": 260,
124
+ "right_thumb_3_joint": 1
125
+ },
126
+ "joint_limit_hard_episode_events": 320,
127
+ "finger_joint_limit_saturation_counts": {
128
+ "left_thumb_2_joint": 64,
129
+ "right_little_1_joint": 35,
130
+ "right_thumb_2_joint": 146
131
+ },
132
+ "joint_limit_max_excess_by_joint": {
133
+ "left_index_1_joint": 0.09537839889526367,
134
+ "left_index_2_joint": 5.733663499540853e-08,
135
+ "left_little_1_joint": 0.006352424621582031,
136
+ "left_little_2_joint": 2.830829615163566e-08,
137
+ "left_middle_1_joint": 0.0744239091873169,
138
+ "left_ring_1_joint": 0.17148566246032715,
139
+ "left_ring_2_joint": 1.3399794340784865e-07,
140
+ "left_thumb_1_joint": 0.003964900970458984,
141
+ "left_thumb_2_joint": 0.32661721110343933,
142
+ "left_thumb_3_joint": 0.003224111394956708,
143
+ "left_thumb_4_joint": 0.0007645785808563232,
144
+ "right_index_1_joint": 0.0020209334325045347,
145
+ "right_index_2_joint": 4.797031124326168e-06,
146
+ "right_little_1_joint": 0.1665497124195099,
147
+ "right_little_2_joint": 2.012267395912204e-05,
148
+ "right_middle_1_joint": 0.16555047035217285,
149
+ "right_middle_2_joint": 6.441100140364142e-06,
150
+ "right_ring_1_joint": 0.003799001919105649,
151
+ "right_ring_2_joint": 8.341282409674022e-06,
152
+ "right_thumb_1_joint": 0.001817941665649414,
153
+ "right_thumb_2_joint": 0.6389230489730835,
154
+ "right_thumb_3_joint": 0.005630647297948599
155
+ },
156
+ "coverage": {
157
+ "hard_limit_source": {
158
+ "available": 25,
159
+ "unavailable": 0,
160
+ "unknown": 0
161
+ },
162
+ "joint_role_manifest": {
163
+ "available": 25,
164
+ "unavailable": 0,
165
+ "unknown": 0
166
+ }
167
+ },
168
+ "hard_limit_sources": {
169
+ "runtime.joint_pos_limits": 25
170
+ },
171
+ "soft_limit_sources": {
172
+ "runtime.soft_joint_pos_limits": 25
173
+ },
174
+ "unknown_runtime_joint_names": [],
175
+ "baseline_checks": {
176
+ "post_home": {
177
+ "checked": 25,
178
+ "hard_violation_episodes": 0
179
+ },
180
+ "post_reset": {
181
+ "checked": 25,
182
+ "hard_violation_episodes": 0
183
+ }
184
+ }
185
+ }
186
+ },
187
+ "robot_motion": {
188
+ "joint_acceleration_rms": 31.81131845080256,
189
+ "joint_effort_rms": 28.400488049002515,
190
+ "joint_jerk_rms": 3262.490953153297,
191
+ "joint_velocity_rms": 0.7812227406527761
192
+ },
193
+ "grasp": {
194
+ "kinematic_grasp_stability_index": 0.7606007962797874,
195
+ "mean_kinematic_grasp_stability": 0.20402904429432714,
196
+ "per_object_best_kinematic_grasp_stability": {
197
+ "obj_011_banana_2": 0.5000967115593967,
198
+ "obj_013_apple_3": 0.6594170479331068,
199
+ "obj_013_apple_4": 0.3236974894110364
200
+ },
201
+ "grasp_gsi_diagnostics": {
202
+ "mean_anti_slip_score": 0.9005696040111125,
203
+ "mean_duration_score": 0.18084953441139212,
204
+ "mean_hold_score": 0.10100892791406334,
205
+ "mean_kinematic_stability_score": 0.04799742503271932,
206
+ "mean_legacy_gsi": 0.2841188595192751,
207
+ "mean_lift_score": 0.20159079336728772,
208
+ "mean_motion_stability_score": 0.10044899625873113
209
+ }
210
+ },
211
+ "tool": {
212
+ "tool_selection_accuracy": null,
213
+ "tool_switch_success_rate": null,
214
+ "tool_switch_total": 0,
215
+ "tool_switch_successful": 0
216
+ },
217
+ "runtime": {
218
+ "avg_steps": 1849.24,
219
+ "error_count": 0
220
+ }
221
+ }
222
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/native_shards/001/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/native_shards/001/summary.json ADDED
@@ -0,0 +1,221 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 100600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 13,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.52,
50
+ "success_rate_ci": {
51
+ "mean": 0.52,
52
+ "ci_low": 0.32,
53
+ "ci_high": 0.72
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.79
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.316666666666226
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.52,
64
+ "hard_violation_rate": 0.0,
65
+ "drop_rate": 0.0,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 0.0
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.6,
74
+ "at_end_success_rate": 0.56,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.52
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.77,
81
+ "current_chain_depth_progress_score": 0.8133333333333334,
82
+ "latched_chain_depth_progress_score": 0.84,
83
+ "avg_current_chain_depth": 2.44,
84
+ "avg_latched_chain_depth": 2.52
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7039070445474388,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1819,
91
+ "avg_policy_steps_to_success": 606.4615384615385,
92
+ "avg_policy_queries_to_success": 606.4615384615385
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 0.0,
96
+ "violation_counts": {},
97
+ "robot_constraint_violation_rate": 1.0,
98
+ "finger_joint_limit_saturation_rate": 0.36,
99
+ "joint_limit_active_step_rate": 0.08610533981622293,
100
+ "finger_joint_limit_saturation_step_rate": 0.0028273395163535887,
101
+ "max_joint_limit_excess_rad": 0.7528724670410156,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_little_1_joint": 18,
105
+ "left_ring_1_joint": 142,
106
+ "left_thumb_2_joint": 67,
107
+ "right_index_1_joint": 19,
108
+ "right_little_1_joint": 623,
109
+ "right_middle_1_joint": 23,
110
+ "right_ring_1_joint": 18,
111
+ "right_thumb_2_joint": 3384
112
+ },
113
+ "joint_limit_hard_counts": {
114
+ "left_little_1_joint": 4,
115
+ "left_ring_1_joint": 17,
116
+ "left_thumb_2_joint": 45,
117
+ "left_thumb_3_joint": 2,
118
+ "right_index_1_joint": 1,
119
+ "right_little_1_joint": 48,
120
+ "right_middle_1_joint": 3,
121
+ "right_ring_1_joint": 1,
122
+ "right_thumb_2_joint": 208,
123
+ "right_thumb_3_joint": 1
124
+ },
125
+ "joint_limit_hard_episode_events": 295,
126
+ "finger_joint_limit_saturation_counts": {
127
+ "left_little_1_joint": 5,
128
+ "left_ring_1_joint": 94,
129
+ "right_little_1_joint": 28,
130
+ "right_middle_1_joint": 1,
131
+ "right_thumb_2_joint": 15
132
+ },
133
+ "joint_limit_max_excess_by_joint": {
134
+ "left_index_2_joint": 5.705875949502115e-08,
135
+ "left_little_1_joint": 0.2990298271179199,
136
+ "left_little_2_joint": 2.6312754641821812e-08,
137
+ "left_ring_1_joint": 0.7528724670410156,
138
+ "left_ring_2_joint": 1.3446806690353696e-07,
139
+ "left_thumb_1_joint": 0.0007226467132568359,
140
+ "left_thumb_2_joint": 0.05989643931388855,
141
+ "left_thumb_3_joint": 0.006721653509885073,
142
+ "left_thumb_4_joint": 0.0009230077266693115,
143
+ "right_index_1_joint": 0.14507833123207092,
144
+ "right_index_2_joint": 5.033112756791525e-06,
145
+ "right_little_1_joint": 0.12037502974271774,
146
+ "right_little_2_joint": 2.96342095680302e-05,
147
+ "right_middle_1_joint": 0.19405332207679749,
148
+ "right_middle_2_joint": 2.9512741548387567e-06,
149
+ "right_ring_1_joint": 0.19054749608039856,
150
+ "right_ring_2_joint": 8.576511390856467e-06,
151
+ "right_thumb_1_joint": 0.0024912357330322266,
152
+ "right_thumb_2_joint": 0.1125117763876915,
153
+ "right_thumb_3_joint": 0.007149942219257355
154
+ },
155
+ "coverage": {
156
+ "hard_limit_source": {
157
+ "available": 25,
158
+ "unavailable": 0,
159
+ "unknown": 0
160
+ },
161
+ "joint_role_manifest": {
162
+ "available": 25,
163
+ "unavailable": 0,
164
+ "unknown": 0
165
+ }
166
+ },
167
+ "hard_limit_sources": {
168
+ "runtime.joint_pos_limits": 25
169
+ },
170
+ "soft_limit_sources": {
171
+ "runtime.soft_joint_pos_limits": 25
172
+ },
173
+ "unknown_runtime_joint_names": [],
174
+ "baseline_checks": {
175
+ "post_home": {
176
+ "checked": 25,
177
+ "hard_violation_episodes": 0
178
+ },
179
+ "post_reset": {
180
+ "checked": 25,
181
+ "hard_violation_episodes": 0
182
+ }
183
+ }
184
+ }
185
+ },
186
+ "robot_motion": {
187
+ "joint_acceleration_rms": 31.87178627152146,
188
+ "joint_effort_rms": 28.638089834672794,
189
+ "joint_jerk_rms": 3271.580743717676,
190
+ "joint_velocity_rms": 0.7875052680631838
191
+ },
192
+ "grasp": {
193
+ "kinematic_grasp_stability_index": 0.7986941344790008,
194
+ "mean_kinematic_grasp_stability": 0.25278086245253467,
195
+ "per_object_best_kinematic_grasp_stability": {
196
+ "obj_011_banana_2": 0.520039709921998,
197
+ "obj_013_apple_3": 0.6584817594997101,
198
+ "obj_013_apple_4": 0.4171289785423111
199
+ },
200
+ "grasp_gsi_diagnostics": {
201
+ "mean_anti_slip_score": 0.9112831621248123,
202
+ "mean_duration_score": 0.19039473706878018,
203
+ "mean_hold_score": 0.1148385657824884,
204
+ "mean_kinematic_stability_score": 0.05929122540572427,
205
+ "mean_legacy_gsi": 0.29739843888080175,
206
+ "mean_lift_score": 0.2152960211041506,
207
+ "mean_motion_stability_score": 0.11781868808324267
208
+ }
209
+ },
210
+ "tool": {
211
+ "tool_selection_accuracy": null,
212
+ "tool_switch_success_rate": null,
213
+ "tool_switch_total": 0,
214
+ "tool_switch_successful": 0
215
+ },
216
+ "runtime": {
217
+ "avg_steps": 1867.48,
218
+ "error_count": 0
219
+ }
220
+ }
221
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Equi/summary.json ADDED
@@ -0,0 +1,231 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 100600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 29,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.58,
50
+ "success_rate_ci": {
51
+ "mean": 0.58,
52
+ "ci_low": 0.44,
53
+ "ci_high": 0.72
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.805
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.228735632183472
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.58,
64
+ "hard_violation_rate": 0.04,
65
+ "drop_rate": 0.04,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 2.1524354807464648e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.62,
74
+ "at_end_success_rate": 0.6,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.58
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.79,
81
+ "current_chain_depth_progress_score": 0.8333333333333334,
82
+ "latched_chain_depth_progress_score": 0.8533333333333333,
83
+ "avg_current_chain_depth": 2.5,
84
+ "avg_latched_chain_depth": 2.56
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7060413074319009,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1813.7241379310344,
91
+ "avg_policy_steps_to_success": 604.8275862068965,
92
+ "avg_policy_queries_to_success": 604.8275862068965
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 2.1524354807464648e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 1,
98
+ "drop_to_table_obj_013_apple_4": 1
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.32,
102
+ "joint_limit_active_step_rate": 0.08905701801588498,
103
+ "finger_joint_limit_saturation_step_rate": 0.0039927678167846915,
104
+ "max_joint_limit_excess_rad": 0.7528724670410156,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "left_index_1_joint": 18,
108
+ "left_little_1_joint": 18,
109
+ "left_middle_1_joint": 6,
110
+ "left_ring_1_joint": 178,
111
+ "left_thumb_2_joint": 174,
112
+ "right_index_1_joint": 19,
113
+ "right_little_1_joint": 1509,
114
+ "right_middle_1_joint": 55,
115
+ "right_ring_1_joint": 18,
116
+ "right_thumb_2_joint": 6740
117
+ },
118
+ "joint_limit_hard_counts": {
119
+ "left_index_1_joint": 1,
120
+ "left_little_1_joint": 9,
121
+ "left_middle_1_joint": 1,
122
+ "left_ring_1_joint": 23,
123
+ "left_thumb_2_joint": 79,
124
+ "left_thumb_3_joint": 2,
125
+ "right_index_1_joint": 1,
126
+ "right_little_1_joint": 91,
127
+ "right_middle_1_joint": 7,
128
+ "right_ring_1_joint": 1,
129
+ "right_thumb_2_joint": 468,
130
+ "right_thumb_3_joint": 2
131
+ },
132
+ "joint_limit_hard_episode_events": 615,
133
+ "finger_joint_limit_saturation_counts": {
134
+ "left_little_1_joint": 5,
135
+ "left_ring_1_joint": 94,
136
+ "left_thumb_2_joint": 64,
137
+ "right_little_1_joint": 63,
138
+ "right_middle_1_joint": 1,
139
+ "right_thumb_2_joint": 161
140
+ },
141
+ "joint_limit_max_excess_by_joint": {
142
+ "left_index_1_joint": 0.09537839889526367,
143
+ "left_index_2_joint": 5.733663499540853e-08,
144
+ "left_little_1_joint": 0.2990298271179199,
145
+ "left_little_2_joint": 2.830829615163566e-08,
146
+ "left_middle_1_joint": 0.0744239091873169,
147
+ "left_ring_1_joint": 0.7528724670410156,
148
+ "left_ring_2_joint": 1.3446806690353696e-07,
149
+ "left_thumb_1_joint": 0.003964900970458984,
150
+ "left_thumb_2_joint": 0.32661721110343933,
151
+ "left_thumb_3_joint": 0.006721653509885073,
152
+ "left_thumb_4_joint": 0.0009230077266693115,
153
+ "right_index_1_joint": 0.14507833123207092,
154
+ "right_index_2_joint": 5.033112756791525e-06,
155
+ "right_little_1_joint": 0.1665497124195099,
156
+ "right_little_2_joint": 2.96342095680302e-05,
157
+ "right_middle_1_joint": 0.19405332207679749,
158
+ "right_middle_2_joint": 6.441100140364142e-06,
159
+ "right_ring_1_joint": 0.19054749608039856,
160
+ "right_ring_2_joint": 8.576511390856467e-06,
161
+ "right_thumb_1_joint": 0.0024912357330322266,
162
+ "right_thumb_2_joint": 0.6389230489730835,
163
+ "right_thumb_3_joint": 0.007149942219257355
164
+ },
165
+ "coverage": {
166
+ "hard_limit_source": {
167
+ "available": 50,
168
+ "unavailable": 0,
169
+ "unknown": 0
170
+ },
171
+ "joint_role_manifest": {
172
+ "available": 50,
173
+ "unavailable": 0,
174
+ "unknown": 0
175
+ }
176
+ },
177
+ "hard_limit_sources": {
178
+ "runtime.joint_pos_limits": 50
179
+ },
180
+ "soft_limit_sources": {
181
+ "runtime.soft_joint_pos_limits": 50
182
+ },
183
+ "unknown_runtime_joint_names": [],
184
+ "baseline_checks": {
185
+ "post_home": {
186
+ "checked": 50,
187
+ "hard_violation_episodes": 0
188
+ },
189
+ "post_reset": {
190
+ "checked": 50,
191
+ "hard_violation_episodes": 0
192
+ }
193
+ }
194
+ }
195
+ },
196
+ "robot_motion": {
197
+ "joint_acceleration_rms": 31.84155236116201,
198
+ "joint_effort_rms": 28.519288941837655,
199
+ "joint_jerk_rms": 3267.0358484354865,
200
+ "joint_velocity_rms": 0.7843640043579799
201
+ },
202
+ "grasp": {
203
+ "kinematic_grasp_stability_index": 0.7796474653793941,
204
+ "mean_kinematic_grasp_stability": 0.2284049533734309,
205
+ "per_object_best_kinematic_grasp_stability": {
206
+ "obj_011_banana_2": 0.5100682107406973,
207
+ "obj_013_apple_3": 0.6589494037164085,
208
+ "obj_013_apple_4": 0.37041323397667375
209
+ },
210
+ "grasp_gsi_diagnostics": {
211
+ "mean_anti_slip_score": 0.9059263830679625,
212
+ "mean_duration_score": 0.18562213574008615,
213
+ "mean_hold_score": 0.10792374684827587,
214
+ "mean_kinematic_stability_score": 0.05364432521922179,
215
+ "mean_legacy_gsi": 0.29075864920003847,
216
+ "mean_lift_score": 0.20844340723571916,
217
+ "mean_motion_stability_score": 0.1091338421709869
218
+ }
219
+ },
220
+ "tool": {
221
+ "tool_selection_accuracy": null,
222
+ "tool_switch_success_rate": null,
223
+ "tool_switch_total": 0,
224
+ "tool_switch_successful": 0
225
+ },
226
+ "runtime": {
227
+ "avg_steps": 1858.36,
228
+ "error_count": 0
229
+ }
230
+ }
231
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/native_shards/000/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/native_shards/000/summary.json ADDED
@@ -0,0 +1,227 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 200600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 14,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.56,
50
+ "success_rate_ci": {
51
+ "mean": 0.56,
52
+ "ci_low": 0.36,
53
+ "ci_high": 0.76
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.82
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.195238095237663
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.56,
64
+ "hard_violation_rate": 0.08,
65
+ "drop_rate": 0.08,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 4.302555718096549e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.6,
74
+ "at_end_success_rate": 0.56,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.56
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.78,
81
+ "current_chain_depth_progress_score": 0.7866666666666666,
82
+ "latched_chain_depth_progress_score": 0.84,
83
+ "avg_current_chain_depth": 2.36,
84
+ "avg_latched_chain_depth": 2.52
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7071072114846997,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1811.7142857142858,
91
+ "avg_policy_steps_to_success": 604.2857142857143,
92
+ "avg_policy_queries_to_success": 604.2857142857143
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 4.302555718096549e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 2
98
+ },
99
+ "robot_constraint_violation_rate": 0.92,
100
+ "finger_joint_limit_saturation_rate": 0.36,
101
+ "joint_limit_active_step_rate": 0.13393855950434558,
102
+ "finger_joint_limit_saturation_step_rate": 0.004022889596420273,
103
+ "max_joint_limit_excess_rad": 0.4412386417388916,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {
106
+ "left_little_1_joint": 4,
107
+ "left_middle_1_joint": 106,
108
+ "left_ring_1_joint": 36,
109
+ "left_thumb_2_joint": 90,
110
+ "right_index_1_joint": 27,
111
+ "right_little_1_joint": 638,
112
+ "right_middle_1_joint": 29,
113
+ "right_ring_1_joint": 26,
114
+ "right_thumb_2_joint": 5486
115
+ },
116
+ "joint_limit_hard_counts": {
117
+ "left_little_1_joint": 5,
118
+ "left_middle_1_joint": 1,
119
+ "left_ring_1_joint": 12,
120
+ "left_thumb_2_joint": 27,
121
+ "right_index_1_joint": 1,
122
+ "right_little_1_joint": 39,
123
+ "right_middle_1_joint": 1,
124
+ "right_ring_1_joint": 1,
125
+ "right_thumb_2_joint": 260,
126
+ "right_thumb_3_joint": 14
127
+ },
128
+ "joint_limit_hard_episode_events": 305,
129
+ "finger_joint_limit_saturation_counts": {
130
+ "left_middle_1_joint": 103,
131
+ "left_ring_1_joint": 8,
132
+ "left_thumb_2_joint": 4,
133
+ "right_index_1_joint": 11,
134
+ "right_little_1_joint": 60,
135
+ "right_middle_1_joint": 12,
136
+ "right_ring_1_joint": 10
137
+ },
138
+ "joint_limit_max_excess_by_joint": {
139
+ "left_index_2_joint": 5.56565353804217e-08,
140
+ "left_little_1_joint": 0.014103055000305176,
141
+ "left_little_2_joint": 2.691614220395877e-08,
142
+ "left_middle_1_joint": 0.4412386417388916,
143
+ "left_ring_1_joint": 0.4229525327682495,
144
+ "left_ring_2_joint": 1.3574378954217536e-07,
145
+ "left_thumb_1_joint": 0.001951456069946289,
146
+ "left_thumb_2_joint": 0.14655572175979614,
147
+ "left_thumb_3_joint": 0.004819095134735107,
148
+ "left_thumb_4_joint": 0.0015417933464050293,
149
+ "right_index_1_joint": 0.2025446891784668,
150
+ "right_index_2_joint": 3.7390987017715815e-06,
151
+ "right_little_1_joint": 0.18915338814258575,
152
+ "right_little_2_joint": 1.66680747497594e-05,
153
+ "right_middle_1_joint": 0.24720749258995056,
154
+ "right_middle_2_joint": 3.7883514778513927e-06,
155
+ "right_ring_1_joint": 0.19600215554237366,
156
+ "right_ring_2_joint": 1.1636860108410474e-05,
157
+ "right_thumb_1_joint": 0.0017305612564086914,
158
+ "right_thumb_2_joint": 0.08675424009561539,
159
+ "right_thumb_3_joint": 0.006728981621563435
160
+ },
161
+ "coverage": {
162
+ "hard_limit_source": {
163
+ "available": 25,
164
+ "unavailable": 0,
165
+ "unknown": 0
166
+ },
167
+ "joint_role_manifest": {
168
+ "available": 25,
169
+ "unavailable": 0,
170
+ "unknown": 0
171
+ }
172
+ },
173
+ "hard_limit_sources": {
174
+ "runtime.joint_pos_limits": 25
175
+ },
176
+ "soft_limit_sources": {
177
+ "runtime.soft_joint_pos_limits": 25
178
+ },
179
+ "unknown_runtime_joint_names": [],
180
+ "baseline_checks": {
181
+ "post_home": {
182
+ "checked": 25,
183
+ "hard_violation_episodes": 0
184
+ },
185
+ "post_reset": {
186
+ "checked": 25,
187
+ "hard_violation_episodes": 0
188
+ }
189
+ }
190
+ }
191
+ },
192
+ "robot_motion": {
193
+ "joint_acceleration_rms": 31.942701182032884,
194
+ "joint_effort_rms": 29.600517276836182,
195
+ "joint_jerk_rms": 3281.581084945716,
196
+ "joint_velocity_rms": 0.8306126814289537
197
+ },
198
+ "grasp": {
199
+ "kinematic_grasp_stability_index": 0.7429419782828929,
200
+ "mean_kinematic_grasp_stability": 0.21011839365589818,
201
+ "per_object_best_kinematic_grasp_stability": {
202
+ "obj_011_banana_2": 0.5198799886575314,
203
+ "obj_013_apple_3": 0.631973030915136,
204
+ "obj_013_apple_4": 0.3919503790693268
205
+ },
206
+ "grasp_gsi_diagnostics": {
207
+ "mean_anti_slip_score": 0.8991238304262589,
208
+ "mean_duration_score": 0.18809960073164697,
209
+ "mean_hold_score": 0.09181271878768628,
210
+ "mean_kinematic_stability_score": 0.043601601663226144,
211
+ "mean_legacy_gsi": 0.2818201441060885,
212
+ "mean_lift_score": 0.20866362957810772,
213
+ "mean_motion_stability_score": 0.09337333655619645
214
+ }
215
+ },
216
+ "tool": {
217
+ "tool_selection_accuracy": null,
218
+ "tool_switch_success_rate": null,
219
+ "tool_switch_total": 0,
220
+ "tool_switch_successful": 0
221
+ },
222
+ "runtime": {
223
+ "avg_steps": 1859.36,
224
+ "error_count": 0
225
+ }
226
+ }
227
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/native_shards/001/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/native_shards/001/summary.json ADDED
@@ -0,0 +1,213 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 200600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 12,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.48,
50
+ "success_rate_ci": {
51
+ "mean": 0.48,
52
+ "ci_low": 0.28,
53
+ "ci_high": 0.68
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.77
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.491666666666216
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.48,
64
+ "hard_violation_rate": 0.16,
65
+ "drop_rate": 0.16,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 8.526239502067613e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.52,
74
+ "at_end_success_rate": 0.48,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.48
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.73,
81
+ "current_chain_depth_progress_score": 0.7466666666666667,
82
+ "latched_chain_depth_progress_score": 0.7999999999999999,
83
+ "avg_current_chain_depth": 2.24,
84
+ "avg_latched_chain_depth": 2.4
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.700361632456088,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1829.5,
91
+ "avg_policy_steps_to_success": 610.1666666666666,
92
+ "avg_policy_queries_to_success": 610.1666666666666
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 8.526239502067613e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 4
98
+ },
99
+ "robot_constraint_violation_rate": 0.88,
100
+ "finger_joint_limit_saturation_rate": 0.32,
101
+ "joint_limit_active_step_rate": 0.10647141578206931,
102
+ "finger_joint_limit_saturation_step_rate": 0.0021315598755169032,
103
+ "max_joint_limit_excess_rad": 0.2878232002258301,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {
106
+ "left_little_1_joint": 14,
107
+ "left_ring_1_joint": 7,
108
+ "left_thumb_2_joint": 105,
109
+ "right_little_1_joint": 569,
110
+ "right_thumb_2_joint": 4580
111
+ },
112
+ "joint_limit_hard_counts": {
113
+ "left_little_1_joint": 1,
114
+ "left_ring_1_joint": 4,
115
+ "left_thumb_2_joint": 27,
116
+ "right_little_1_joint": 35,
117
+ "right_thumb_2_joint": 169
118
+ },
119
+ "joint_limit_hard_episode_events": 211,
120
+ "finger_joint_limit_saturation_counts": {
121
+ "left_little_1_joint": 4,
122
+ "right_little_1_joint": 88,
123
+ "right_thumb_2_joint": 8
124
+ },
125
+ "joint_limit_max_excess_by_joint": {
126
+ "left_index_2_joint": 5.580874073984887e-08,
127
+ "left_little_1_joint": 0.2878232002258301,
128
+ "left_little_2_joint": 2.6108798678592393e-08,
129
+ "left_ring_1_joint": 0.05118584632873535,
130
+ "left_ring_2_joint": 1.3324738290521054e-07,
131
+ "left_thumb_1_joint": 0.002151012420654297,
132
+ "left_thumb_2_joint": 0.11367978900671005,
133
+ "left_thumb_3_joint": 0.0032891035079956055,
134
+ "left_thumb_4_joint": 0.0008653402328491211,
135
+ "right_index_1_joint": 0.0018417403334751725,
136
+ "right_index_2_joint": 5.222084837441798e-06,
137
+ "right_little_1_joint": 0.1569756716489792,
138
+ "right_little_2_joint": 1.4976092643337324e-05,
139
+ "right_middle_1_joint": 0.004284973256289959,
140
+ "right_middle_2_joint": 5.64583115192363e-06,
141
+ "right_ring_1_joint": 0.0018691126024350524,
142
+ "right_ring_2_joint": 4.785727924172534e-06,
143
+ "right_thumb_1_joint": 0.0015932321548461914,
144
+ "right_thumb_2_joint": 0.11849448829889297,
145
+ "right_thumb_3_joint": 0.0036524867173284292
146
+ },
147
+ "coverage": {
148
+ "hard_limit_source": {
149
+ "available": 25,
150
+ "unavailable": 0,
151
+ "unknown": 0
152
+ },
153
+ "joint_role_manifest": {
154
+ "available": 25,
155
+ "unavailable": 0,
156
+ "unknown": 0
157
+ }
158
+ },
159
+ "hard_limit_sources": {
160
+ "runtime.joint_pos_limits": 25
161
+ },
162
+ "soft_limit_sources": {
163
+ "runtime.soft_joint_pos_limits": 25
164
+ },
165
+ "unknown_runtime_joint_names": [],
166
+ "baseline_checks": {
167
+ "post_home": {
168
+ "checked": 25,
169
+ "hard_violation_episodes": 0
170
+ },
171
+ "post_reset": {
172
+ "checked": 25,
173
+ "hard_violation_episodes": 0
174
+ }
175
+ }
176
+ }
177
+ },
178
+ "robot_motion": {
179
+ "joint_acceleration_rms": 30.927881943286327,
180
+ "joint_effort_rms": 28.437855397164057,
181
+ "joint_jerk_rms": 3176.6394146056473,
182
+ "joint_velocity_rms": 0.7888525777349991
183
+ },
184
+ "grasp": {
185
+ "kinematic_grasp_stability_index": 0.7547786333349686,
186
+ "mean_kinematic_grasp_stability": 0.2673630810297875,
187
+ "per_object_best_kinematic_grasp_stability": {
188
+ "obj_011_banana_2": 0.4592384247761509,
189
+ "obj_013_apple_3": 0.622399446704726,
190
+ "obj_013_apple_4": 0.3730962071317398
191
+ },
192
+ "grasp_gsi_diagnostics": {
193
+ "mean_anti_slip_score": 0.8998900102373971,
194
+ "mean_duration_score": 0.1811182823342471,
195
+ "mean_hold_score": 0.10845242194316589,
196
+ "mean_kinematic_stability_score": 0.0658419767766158,
197
+ "mean_legacy_gsi": 0.2883339568502629,
198
+ "mean_lift_score": 0.21004522003626003,
199
+ "mean_motion_stability_score": 0.10859270313137316
200
+ }
201
+ },
202
+ "tool": {
203
+ "tool_selection_accuracy": null,
204
+ "tool_switch_success_rate": null,
205
+ "tool_switch_total": 0,
206
+ "tool_switch_successful": 0
207
+ },
208
+ "runtime": {
209
+ "avg_steps": 1876.56,
210
+ "error_count": 0
211
+ }
212
+ }
213
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Full/summary.json ADDED
@@ -0,0 +1,229 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 200600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 26,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.52,
50
+ "success_rate_ci": {
51
+ "mean": 0.52,
52
+ "ci_low": 0.38,
53
+ "ci_high": 0.66
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.795
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.33205128205084
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.52,
64
+ "hard_violation_rate": 0.12,
65
+ "drop_rate": 0.12,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 6.424120430844344e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.56,
74
+ "at_end_success_rate": 0.52,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.52
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.755,
81
+ "current_chain_depth_progress_score": 0.7666666666666666,
82
+ "latched_chain_depth_progress_score": 0.82,
83
+ "avg_current_chain_depth": 2.3,
84
+ "avg_latched_chain_depth": 2.46
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7039938673176481,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1819.923076923077,
91
+ "avg_policy_steps_to_success": 607,
92
+ "avg_policy_queries_to_success": 607
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 6.424120430844344e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 6
98
+ },
99
+ "robot_constraint_violation_rate": 0.9,
100
+ "finger_joint_limit_saturation_rate": 0.34,
101
+ "joint_limit_active_step_rate": 0.12014175892417396,
102
+ "finger_joint_limit_saturation_step_rate": 0.0030728709394205445,
103
+ "max_joint_limit_excess_rad": 0.4412386417388916,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {
106
+ "left_little_1_joint": 18,
107
+ "left_middle_1_joint": 106,
108
+ "left_ring_1_joint": 43,
109
+ "left_thumb_2_joint": 195,
110
+ "right_index_1_joint": 27,
111
+ "right_little_1_joint": 1207,
112
+ "right_middle_1_joint": 29,
113
+ "right_ring_1_joint": 26,
114
+ "right_thumb_2_joint": 10066
115
+ },
116
+ "joint_limit_hard_counts": {
117
+ "left_little_1_joint": 6,
118
+ "left_middle_1_joint": 1,
119
+ "left_ring_1_joint": 16,
120
+ "left_thumb_2_joint": 54,
121
+ "right_index_1_joint": 1,
122
+ "right_little_1_joint": 74,
123
+ "right_middle_1_joint": 1,
124
+ "right_ring_1_joint": 1,
125
+ "right_thumb_2_joint": 429,
126
+ "right_thumb_3_joint": 14
127
+ },
128
+ "joint_limit_hard_episode_events": 516,
129
+ "finger_joint_limit_saturation_counts": {
130
+ "left_little_1_joint": 4,
131
+ "left_middle_1_joint": 103,
132
+ "left_ring_1_joint": 8,
133
+ "left_thumb_2_joint": 4,
134
+ "right_index_1_joint": 11,
135
+ "right_little_1_joint": 148,
136
+ "right_middle_1_joint": 12,
137
+ "right_ring_1_joint": 10,
138
+ "right_thumb_2_joint": 8
139
+ },
140
+ "joint_limit_max_excess_by_joint": {
141
+ "left_index_2_joint": 5.580874073984887e-08,
142
+ "left_little_1_joint": 0.2878232002258301,
143
+ "left_little_2_joint": 2.691614220395877e-08,
144
+ "left_middle_1_joint": 0.4412386417388916,
145
+ "left_ring_1_joint": 0.4229525327682495,
146
+ "left_ring_2_joint": 1.3574378954217536e-07,
147
+ "left_thumb_1_joint": 0.002151012420654297,
148
+ "left_thumb_2_joint": 0.14655572175979614,
149
+ "left_thumb_3_joint": 0.004819095134735107,
150
+ "left_thumb_4_joint": 0.0015417933464050293,
151
+ "right_index_1_joint": 0.2025446891784668,
152
+ "right_index_2_joint": 5.222084837441798e-06,
153
+ "right_little_1_joint": 0.18915338814258575,
154
+ "right_little_2_joint": 1.66680747497594e-05,
155
+ "right_middle_1_joint": 0.24720749258995056,
156
+ "right_middle_2_joint": 5.64583115192363e-06,
157
+ "right_ring_1_joint": 0.19600215554237366,
158
+ "right_ring_2_joint": 1.1636860108410474e-05,
159
+ "right_thumb_1_joint": 0.0017305612564086914,
160
+ "right_thumb_2_joint": 0.11849448829889297,
161
+ "right_thumb_3_joint": 0.006728981621563435
162
+ },
163
+ "coverage": {
164
+ "hard_limit_source": {
165
+ "available": 50,
166
+ "unavailable": 0,
167
+ "unknown": 0
168
+ },
169
+ "joint_role_manifest": {
170
+ "available": 50,
171
+ "unavailable": 0,
172
+ "unknown": 0
173
+ }
174
+ },
175
+ "hard_limit_sources": {
176
+ "runtime.joint_pos_limits": 50
177
+ },
178
+ "soft_limit_sources": {
179
+ "runtime.soft_joint_pos_limits": 50
180
+ },
181
+ "unknown_runtime_joint_names": [],
182
+ "baseline_checks": {
183
+ "post_home": {
184
+ "checked": 50,
185
+ "hard_violation_episodes": 0
186
+ },
187
+ "post_reset": {
188
+ "checked": 50,
189
+ "hard_violation_episodes": 0
190
+ }
191
+ }
192
+ }
193
+ },
194
+ "robot_motion": {
195
+ "joint_acceleration_rms": 31.435291562659604,
196
+ "joint_effort_rms": 29.01918633700012,
197
+ "joint_jerk_rms": 3229.110249775682,
198
+ "joint_velocity_rms": 0.8097326295819764
199
+ },
200
+ "grasp": {
201
+ "kinematic_grasp_stability_index": 0.7488603058089308,
202
+ "mean_kinematic_grasp_stability": 0.23874073734284285,
203
+ "per_object_best_kinematic_grasp_stability": {
204
+ "obj_011_banana_2": 0.4895592067168411,
205
+ "obj_013_apple_3": 0.627186238809931,
206
+ "obj_013_apple_4": 0.3825232931005333
207
+ },
208
+ "grasp_gsi_diagnostics": {
209
+ "mean_anti_slip_score": 0.899506920331828,
210
+ "mean_duration_score": 0.18460894153294702,
211
+ "mean_hold_score": 0.10013257036542608,
212
+ "mean_kinematic_stability_score": 0.05472178921992097,
213
+ "mean_legacy_gsi": 0.2850770504781757,
214
+ "mean_lift_score": 0.2093544248071839,
215
+ "mean_motion_stability_score": 0.10098301984378481
216
+ }
217
+ },
218
+ "tool": {
219
+ "tool_selection_accuracy": null,
220
+ "tool_switch_success_rate": null,
221
+ "tool_switch_total": 0,
222
+ "tool_switch_successful": 0
223
+ },
224
+ "runtime": {
225
+ "avg_steps": 1867.96,
226
+ "error_count": 0
227
+ }
228
+ }
229
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/native_shards/000/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/native_shards/000/summary.json ADDED
@@ -0,0 +1,214 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 100600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 7,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.28,
50
+ "success_rate_ci": {
51
+ "mean": 0.28,
52
+ "ci_low": 0.12,
53
+ "ci_high": 0.48
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.78
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.783333333332866
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.28,
64
+ "hard_violation_rate": 0.36,
65
+ "drop_rate": 0.36,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 0.00018951757249047147
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.36,
74
+ "at_end_success_rate": 0.28,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.28
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.75,
81
+ "current_chain_depth_progress_score": 0.7333333333333333,
82
+ "latched_chain_depth_progress_score": 0.7733333333333333,
83
+ "avg_current_chain_depth": 2.2,
84
+ "avg_latched_chain_depth": 2.32
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.6934073502870508,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1847,
91
+ "avg_policy_steps_to_success": 616,
92
+ "avg_policy_queries_to_success": 616
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 0.00018951757249047147,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 8,
98
+ "drop_to_table_obj_013_apple_3": 1
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.2,
102
+ "joint_limit_active_step_rate": 0.09924403546084357,
103
+ "finger_joint_limit_saturation_step_rate": 0.0009475878624523574,
104
+ "max_joint_limit_excess_rad": 0.19214797019958496,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "left_index_1_joint": 14,
108
+ "left_little_1_joint": 1,
109
+ "left_ring_1_joint": 17,
110
+ "left_thumb_2_joint": 44,
111
+ "right_little_1_joint": 758,
112
+ "right_thumb_2_joint": 3997
113
+ },
114
+ "joint_limit_hard_counts": {
115
+ "left_index_1_joint": 1,
116
+ "left_little_1_joint": 1,
117
+ "left_ring_1_joint": 4,
118
+ "left_thumb_2_joint": 31,
119
+ "right_little_1_joint": 50,
120
+ "right_thumb_2_joint": 195
121
+ },
122
+ "joint_limit_hard_episode_events": 255,
123
+ "finger_joint_limit_saturation_counts": {
124
+ "right_little_1_joint": 45
125
+ },
126
+ "joint_limit_max_excess_by_joint": {
127
+ "left_index_1_joint": 0.16644299030303955,
128
+ "left_index_2_joint": 5.658005974851221e-08,
129
+ "left_little_1_joint": 0.016172051429748535,
130
+ "left_little_2_joint": 2.5896374822309554e-08,
131
+ "left_ring_1_joint": 0.07299530506134033,
132
+ "left_ring_2_joint": 1.3329787407201366e-07,
133
+ "left_thumb_1_joint": 0.0016608238220214844,
134
+ "left_thumb_2_joint": 0.0968337282538414,
135
+ "left_thumb_3_joint": 0.0030185580253601074,
136
+ "left_thumb_4_joint": 0.00218924880027771,
137
+ "right_index_2_joint": 4.040296062157722e-06,
138
+ "right_little_1_joint": 0.19214797019958496,
139
+ "right_little_2_joint": 1.838919342844747e-05,
140
+ "right_middle_1_joint": 0.004355231765657663,
141
+ "right_middle_2_joint": 4.5347633204073645e-06,
142
+ "right_ring_1_joint": 0.0027330464217811823,
143
+ "right_ring_2_joint": 1.2389589755912311e-05,
144
+ "right_thumb_1_joint": 0.001923680305480957,
145
+ "right_thumb_2_joint": 0.09849416464567184,
146
+ "right_thumb_3_joint": 0.004501382354646921
147
+ },
148
+ "coverage": {
149
+ "hard_limit_source": {
150
+ "available": 25,
151
+ "unavailable": 0,
152
+ "unknown": 0
153
+ },
154
+ "joint_role_manifest": {
155
+ "available": 25,
156
+ "unavailable": 0,
157
+ "unknown": 0
158
+ }
159
+ },
160
+ "hard_limit_sources": {
161
+ "runtime.joint_pos_limits": 25
162
+ },
163
+ "soft_limit_sources": {
164
+ "runtime.soft_joint_pos_limits": 25
165
+ },
166
+ "unknown_runtime_joint_names": [],
167
+ "baseline_checks": {
168
+ "post_home": {
169
+ "checked": 25,
170
+ "hard_violation_episodes": 0
171
+ },
172
+ "post_reset": {
173
+ "checked": 25,
174
+ "hard_violation_episodes": 0
175
+ }
176
+ }
177
+ }
178
+ },
179
+ "robot_motion": {
180
+ "joint_acceleration_rms": 31.752812001535982,
181
+ "joint_effort_rms": 28.87644765283747,
182
+ "joint_jerk_rms": 3259.5732690363793,
183
+ "joint_velocity_rms": 0.798696026998799
184
+ },
185
+ "grasp": {
186
+ "kinematic_grasp_stability_index": 0.8258457927407302,
187
+ "mean_kinematic_grasp_stability": 0.2827305907195098,
188
+ "per_object_best_kinematic_grasp_stability": {
189
+ "obj_011_banana_2": 0.4102134107387316,
190
+ "obj_013_apple_3": 0.74476144979974,
191
+ "obj_013_apple_4": 0.36968266414967843
192
+ },
193
+ "grasp_gsi_diagnostics": {
194
+ "mean_anti_slip_score": 0.9019582214792651,
195
+ "mean_duration_score": 0.2049256981850579,
196
+ "mean_hold_score": 0.13677794547516978,
197
+ "mean_kinematic_stability_score": 0.07685980941425512,
198
+ "mean_legacy_gsi": 0.31140311078536226,
199
+ "mean_lift_score": 0.245007537179553,
200
+ "mean_motion_stability_score": 0.13722380195035103
201
+ }
202
+ },
203
+ "tool": {
204
+ "tool_selection_accuracy": null,
205
+ "tool_switch_success_rate": null,
206
+ "tool_switch_total": 0,
207
+ "tool_switch_successful": 0
208
+ },
209
+ "runtime": {
210
+ "avg_steps": 1899.56,
211
+ "error_count": 0
212
+ }
213
+ }
214
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/native_shards/001/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/native_shards/001/summary.json ADDED
@@ -0,0 +1,216 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 100600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 14,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.56,
50
+ "success_rate_ci": {
51
+ "mean": 0.56,
52
+ "ci_low": 0.36,
53
+ "ci_high": 0.76
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.79
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.10595238095195
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.56,
64
+ "hard_violation_rate": 0.12,
65
+ "drop_rate": 0.12,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 6.46426339718589e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.6,
74
+ "at_end_success_rate": 0.56,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.56
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.78,
81
+ "current_chain_depth_progress_score": 0.8133333333333334,
82
+ "latched_chain_depth_progress_score": 0.8266666666666667,
83
+ "avg_current_chain_depth": 2.44,
84
+ "avg_latched_chain_depth": 2.48
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7090558529241183,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1806.357142857143,
91
+ "avg_policy_steps_to_success": 602.5714285714286,
92
+ "avg_policy_queries_to_success": 602.5714285714286
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 6.46426339718589e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 3
98
+ },
99
+ "robot_constraint_violation_rate": 1.0,
100
+ "finger_joint_limit_saturation_rate": 0.16,
101
+ "joint_limit_active_step_rate": 0.09028421211402961,
102
+ "finger_joint_limit_saturation_step_rate": 0.0017022560279256178,
103
+ "max_joint_limit_excess_rad": 0.22199881076812744,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {
106
+ "left_ring_1_joint": 102,
107
+ "left_thumb_2_joint": 87,
108
+ "right_little_1_joint": 689,
109
+ "right_middle_1_joint": 17,
110
+ "right_thumb_2_joint": 3485
111
+ },
112
+ "joint_limit_hard_counts": {
113
+ "left_little_1_joint": 1,
114
+ "left_ring_1_joint": 12,
115
+ "left_thumb_2_joint": 25,
116
+ "left_thumb_3_joint": 1,
117
+ "right_little_1_joint": 33,
118
+ "right_middle_1_joint": 1,
119
+ "right_thumb_2_joint": 263,
120
+ "right_thumb_3_joint": 2
121
+ },
122
+ "joint_limit_hard_episode_events": 310,
123
+ "finger_joint_limit_saturation_counts": {
124
+ "left_ring_1_joint": 25,
125
+ "left_thumb_2_joint": 3,
126
+ "right_little_1_joint": 51
127
+ },
128
+ "joint_limit_max_excess_by_joint": {
129
+ "left_index_2_joint": 5.533728142381733e-08,
130
+ "left_little_1_joint": 0.006698250770568848,
131
+ "left_little_2_joint": 2.830829615163566e-08,
132
+ "left_ring_1_joint": 0.22199881076812744,
133
+ "left_ring_2_joint": 1.3474911497723951e-07,
134
+ "left_thumb_1_joint": 0.001912236213684082,
135
+ "left_thumb_2_joint": 0.12226694822311401,
136
+ "left_thumb_3_joint": 0.0055075883865356445,
137
+ "left_thumb_4_joint": 0.001256793737411499,
138
+ "right_index_1_joint": 0.0002675346622709185,
139
+ "right_index_2_joint": 5.087274075776804e-06,
140
+ "right_little_1_joint": 0.19007107615470886,
141
+ "right_little_2_joint": 1.652433093113359e-05,
142
+ "right_middle_1_joint": 0.04379422962665558,
143
+ "right_middle_2_joint": 3.963803465012461e-06,
144
+ "right_ring_1_joint": 0.00020664084877353162,
145
+ "right_ring_2_joint": 1.4466636457655113e-05,
146
+ "right_thumb_1_joint": 0.001809835433959961,
147
+ "right_thumb_2_joint": 0.10138297080993652,
148
+ "right_thumb_3_joint": 0.0057983845472335815
149
+ },
150
+ "coverage": {
151
+ "hard_limit_source": {
152
+ "available": 25,
153
+ "unavailable": 0,
154
+ "unknown": 0
155
+ },
156
+ "joint_role_manifest": {
157
+ "available": 25,
158
+ "unavailable": 0,
159
+ "unknown": 0
160
+ }
161
+ },
162
+ "hard_limit_sources": {
163
+ "runtime.joint_pos_limits": 25
164
+ },
165
+ "soft_limit_sources": {
166
+ "runtime.soft_joint_pos_limits": 25
167
+ },
168
+ "unknown_runtime_joint_names": [],
169
+ "baseline_checks": {
170
+ "post_home": {
171
+ "checked": 25,
172
+ "hard_violation_episodes": 0
173
+ },
174
+ "post_reset": {
175
+ "checked": 25,
176
+ "hard_violation_episodes": 0
177
+ }
178
+ }
179
+ }
180
+ },
181
+ "robot_motion": {
182
+ "joint_acceleration_rms": 31.027050901615876,
183
+ "joint_effort_rms": 28.0782669866655,
184
+ "joint_jerk_rms": 3172.0252391728773,
185
+ "joint_velocity_rms": 0.7739043259413048
186
+ },
187
+ "grasp": {
188
+ "kinematic_grasp_stability_index": 0.7928013863762072,
189
+ "mean_kinematic_grasp_stability": 0.2839399731964235,
190
+ "per_object_best_kinematic_grasp_stability": {
191
+ "obj_011_banana_2": 0.49105019200660804,
192
+ "obj_013_apple_3": 0.6302293774685349,
193
+ "obj_013_apple_4": 0.2686502654283713
194
+ },
195
+ "grasp_gsi_diagnostics": {
196
+ "mean_anti_slip_score": 0.8782702021270168,
197
+ "mean_duration_score": 0.212825952326114,
198
+ "mean_hold_score": 0.11642353213340946,
199
+ "mean_kinematic_stability_score": 0.062433499270046794,
200
+ "mean_legacy_gsi": 0.291584832541053,
201
+ "mean_lift_score": 0.2009926321606389,
202
+ "mean_motion_stability_score": 0.11039171614890825
203
+ }
204
+ },
205
+ "tool": {
206
+ "tool_selection_accuracy": null,
207
+ "tool_switch_success_rate": null,
208
+ "tool_switch_total": 0,
209
+ "tool_switch_successful": 0
210
+ },
211
+ "runtime": {
212
+ "avg_steps": 1856.36,
213
+ "error_count": 0
214
+ }
215
+ }
216
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/Inv/summary.json ADDED
@@ -0,0 +1,221 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 100600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 21,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.42,
50
+ "success_rate_ci": {
51
+ "mean": 0.42,
52
+ "ci_low": 0.28,
53
+ "ci_high": 0.56
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.785
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.331746031745592
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.42,
64
+ "hard_violation_rate": 0.24,
65
+ "drop_rate": 0.24,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 0.00012779824916398647
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.48,
74
+ "at_end_success_rate": 0.42,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.42
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.765,
81
+ "current_chain_depth_progress_score": 0.7733333333333333,
82
+ "latched_chain_depth_progress_score": 0.7999999999999999,
83
+ "avg_current_chain_depth": 2.32,
84
+ "avg_latched_chain_depth": 2.4
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7038396853784292,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1819.904761904762,
91
+ "avg_policy_steps_to_success": 607.047619047619,
92
+ "avg_policy_queries_to_success": 607.047619047619
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 0.00012779824916398647,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 11,
98
+ "drop_to_table_obj_013_apple_3": 1
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.18,
102
+ "joint_limit_active_step_rate": 0.09481565102558095,
103
+ "finger_joint_limit_saturation_step_rate": 0.00132058190802786,
104
+ "max_joint_limit_excess_rad": 0.22199881076812744,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "left_index_1_joint": 14,
108
+ "left_little_1_joint": 1,
109
+ "left_ring_1_joint": 119,
110
+ "left_thumb_2_joint": 131,
111
+ "right_little_1_joint": 1447,
112
+ "right_middle_1_joint": 17,
113
+ "right_thumb_2_joint": 7482
114
+ },
115
+ "joint_limit_hard_counts": {
116
+ "left_index_1_joint": 1,
117
+ "left_little_1_joint": 2,
118
+ "left_ring_1_joint": 16,
119
+ "left_thumb_2_joint": 56,
120
+ "left_thumb_3_joint": 1,
121
+ "right_little_1_joint": 83,
122
+ "right_middle_1_joint": 1,
123
+ "right_thumb_2_joint": 458,
124
+ "right_thumb_3_joint": 2
125
+ },
126
+ "joint_limit_hard_episode_events": 565,
127
+ "finger_joint_limit_saturation_counts": {
128
+ "left_ring_1_joint": 25,
129
+ "left_thumb_2_joint": 3,
130
+ "right_little_1_joint": 96
131
+ },
132
+ "joint_limit_max_excess_by_joint": {
133
+ "left_index_1_joint": 0.16644299030303955,
134
+ "left_index_2_joint": 5.658005974851221e-08,
135
+ "left_little_1_joint": 0.016172051429748535,
136
+ "left_little_2_joint": 2.830829615163566e-08,
137
+ "left_ring_1_joint": 0.22199881076812744,
138
+ "left_ring_2_joint": 1.3474911497723951e-07,
139
+ "left_thumb_1_joint": 0.001912236213684082,
140
+ "left_thumb_2_joint": 0.12226694822311401,
141
+ "left_thumb_3_joint": 0.0055075883865356445,
142
+ "left_thumb_4_joint": 0.00218924880027771,
143
+ "right_index_1_joint": 0.0002675346622709185,
144
+ "right_index_2_joint": 5.087274075776804e-06,
145
+ "right_little_1_joint": 0.19214797019958496,
146
+ "right_little_2_joint": 1.838919342844747e-05,
147
+ "right_middle_1_joint": 0.04379422962665558,
148
+ "right_middle_2_joint": 4.5347633204073645e-06,
149
+ "right_ring_1_joint": 0.0027330464217811823,
150
+ "right_ring_2_joint": 1.4466636457655113e-05,
151
+ "right_thumb_1_joint": 0.001923680305480957,
152
+ "right_thumb_2_joint": 0.10138297080993652,
153
+ "right_thumb_3_joint": 0.0057983845472335815
154
+ },
155
+ "coverage": {
156
+ "hard_limit_source": {
157
+ "available": 50,
158
+ "unavailable": 0,
159
+ "unknown": 0
160
+ },
161
+ "joint_role_manifest": {
162
+ "available": 50,
163
+ "unavailable": 0,
164
+ "unknown": 0
165
+ }
166
+ },
167
+ "hard_limit_sources": {
168
+ "runtime.joint_pos_limits": 50
169
+ },
170
+ "soft_limit_sources": {
171
+ "runtime.soft_joint_pos_limits": 50
172
+ },
173
+ "unknown_runtime_joint_names": [],
174
+ "baseline_checks": {
175
+ "post_home": {
176
+ "checked": 50,
177
+ "hard_violation_episodes": 0
178
+ },
179
+ "post_reset": {
180
+ "checked": 50,
181
+ "hard_violation_episodes": 0
182
+ }
183
+ }
184
+ }
185
+ },
186
+ "robot_motion": {
187
+ "joint_acceleration_rms": 31.38993145157593,
188
+ "joint_effort_rms": 28.477357319751487,
189
+ "joint_jerk_rms": 3215.7992541046283,
190
+ "joint_velocity_rms": 0.7863001764700519
191
+ },
192
+ "grasp": {
193
+ "kinematic_grasp_stability_index": 0.8093235895584687,
194
+ "mean_kinematic_grasp_stability": 0.28333528195796664,
195
+ "per_object_best_kinematic_grasp_stability": {
196
+ "obj_011_banana_2": 0.45063180137266984,
197
+ "obj_013_apple_3": 0.6874954136341375,
198
+ "obj_013_apple_4": 0.3191664647890249
199
+ },
200
+ "grasp_gsi_diagnostics": {
201
+ "mean_anti_slip_score": 0.890114211803141,
202
+ "mean_duration_score": 0.20887582525558596,
203
+ "mean_hold_score": 0.12660073880428963,
204
+ "mean_kinematic_stability_score": 0.06964665434215096,
205
+ "mean_legacy_gsi": 0.3014939716632076,
206
+ "mean_lift_score": 0.22300008467009594,
207
+ "mean_motion_stability_score": 0.12380775904962964
208
+ }
209
+ },
210
+ "tool": {
211
+ "tool_selection_accuracy": null,
212
+ "tool_switch_success_rate": null,
213
+ "tool_switch_total": 0,
214
+ "tool_switch_successful": 0
215
+ },
216
+ "runtime": {
217
+ "avg_steps": 1877.96,
218
+ "error_count": 0
219
+ }
220
+ }
221
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/native_shards/000/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/native_shards/000/summary.json ADDED
@@ -0,0 +1,225 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 100600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 16,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.64,
50
+ "success_rate_ci": {
51
+ "mean": 0.64,
52
+ "ci_low": 0.44,
53
+ "ci_high": 0.8
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.93
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.322916666666224
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.6,
64
+ "hard_violation_rate": 0.16,
65
+ "drop_rate": 0.12,
66
+ "high_speed_violation_rate": 0.04,
67
+ "safety_violation_step_rate": 8.622547962923044e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.76,
74
+ "at_end_success_rate": 0.64,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.64
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.9,
81
+ "current_chain_depth_progress_score": 0.88,
82
+ "latched_chain_depth_progress_score": 0.92,
83
+ "avg_current_chain_depth": 2.64,
84
+ "avg_latched_chain_depth": 2.76
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7041080978291392,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1819.375,
91
+ "avg_policy_steps_to_success": 606.875,
92
+ "avg_policy_queries_to_success": 606.875
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 8.622547962923044e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 3,
98
+ "high_speed_obj_024_bowl_1": 1
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.32,
102
+ "joint_limit_active_step_rate": 0.08193576201767622,
103
+ "finger_joint_limit_saturation_step_rate": 0.002478982539340375,
104
+ "max_joint_limit_excess_rad": 0.2602928578853607,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "left_little_1_joint": 1,
108
+ "left_ring_1_joint": 71,
109
+ "left_thumb_2_joint": 135,
110
+ "right_index_1_joint": 62,
111
+ "right_little_1_joint": 925,
112
+ "right_middle_1_joint": 66,
113
+ "right_ring_1_joint": 61,
114
+ "right_thumb_2_joint": 2835
115
+ },
116
+ "joint_limit_hard_counts": {
117
+ "left_little_1_joint": 4,
118
+ "left_ring_1_joint": 21,
119
+ "left_thumb_2_joint": 35,
120
+ "right_index_1_joint": 2,
121
+ "right_little_1_joint": 49,
122
+ "right_middle_1_joint": 5,
123
+ "right_ring_1_joint": 2,
124
+ "right_thumb_2_joint": 227,
125
+ "right_thumb_3_joint": 2
126
+ },
127
+ "joint_limit_hard_episode_events": 301,
128
+ "finger_joint_limit_saturation_counts": {
129
+ "left_ring_1_joint": 23,
130
+ "left_thumb_2_joint": 7,
131
+ "right_index_1_joint": 28,
132
+ "right_little_1_joint": 38,
133
+ "right_middle_1_joint": 31,
134
+ "right_ring_1_joint": 33,
135
+ "right_thumb_2_joint": 12
136
+ },
137
+ "joint_limit_max_excess_by_joint": {
138
+ "left_index_2_joint": 5.587289564346065e-08,
139
+ "left_little_1_joint": 0.011418342590332031,
140
+ "left_little_2_joint": 2.727411363423471e-08,
141
+ "left_ring_1_joint": 0.18063080310821533,
142
+ "left_ring_2_joint": 1.3556625333421835e-07,
143
+ "left_thumb_1_joint": 0.0017594099044799805,
144
+ "left_thumb_2_joint": 0.11521251499652863,
145
+ "left_thumb_3_joint": 0.004622459411621094,
146
+ "left_thumb_4_joint": 0.001858144998550415,
147
+ "right_index_1_joint": 0.21233883500099182,
148
+ "right_index_2_joint": 5.520537797565339e-06,
149
+ "right_little_1_joint": 0.15717044472694397,
150
+ "right_little_2_joint": 1.4642590940638911e-05,
151
+ "right_middle_1_joint": 0.24825574457645416,
152
+ "right_middle_2_joint": 1.0853015737666283e-05,
153
+ "right_ring_1_joint": 0.2602928578853607,
154
+ "right_ring_2_joint": 1.1067111699958332e-05,
155
+ "right_thumb_1_joint": 0.0015197992324829102,
156
+ "right_thumb_2_joint": 0.09043316543102264,
157
+ "right_thumb_3_joint": 0.006209963466972113
158
+ },
159
+ "coverage": {
160
+ "hard_limit_source": {
161
+ "available": 25,
162
+ "unavailable": 0,
163
+ "unknown": 0
164
+ },
165
+ "joint_role_manifest": {
166
+ "available": 25,
167
+ "unavailable": 0,
168
+ "unknown": 0
169
+ }
170
+ },
171
+ "hard_limit_sources": {
172
+ "runtime.joint_pos_limits": 25
173
+ },
174
+ "soft_limit_sources": {
175
+ "runtime.soft_joint_pos_limits": 25
176
+ },
177
+ "unknown_runtime_joint_names": [],
178
+ "baseline_checks": {
179
+ "post_home": {
180
+ "checked": 25,
181
+ "hard_violation_episodes": 0
182
+ },
183
+ "post_reset": {
184
+ "checked": 25,
185
+ "hard_violation_episodes": 0
186
+ }
187
+ }
188
+ }
189
+ },
190
+ "robot_motion": {
191
+ "joint_acceleration_rms": 34.95352328381145,
192
+ "joint_effort_rms": 30.14491573970212,
193
+ "joint_jerk_rms": 3576.6293281494727,
194
+ "joint_velocity_rms": 0.8596825733214984
195
+ },
196
+ "grasp": {
197
+ "kinematic_grasp_stability_index": 0.8805828791589503,
198
+ "mean_kinematic_grasp_stability": 0.24459588674941668,
199
+ "per_object_best_kinematic_grasp_stability": {
200
+ "obj_011_banana_2": 0.6510661291893157,
201
+ "obj_013_apple_3": 0.752953025482869,
202
+ "obj_013_apple_4": 0.3599093291506452
203
+ },
204
+ "grasp_gsi_diagnostics": {
205
+ "mean_anti_slip_score": 0.8672197469891688,
206
+ "mean_duration_score": 0.223422011725421,
207
+ "mean_hold_score": 0.1359910510889629,
208
+ "mean_kinematic_stability_score": 0.06078691132979986,
209
+ "mean_legacy_gsi": 0.30799883443097015,
210
+ "mean_lift_score": 0.24950608076176442,
211
+ "mean_motion_stability_score": 0.1354889478083832
212
+ }
213
+ },
214
+ "tool": {
215
+ "tool_selection_accuracy": null,
216
+ "tool_switch_success_rate": null,
217
+ "tool_switch_total": 0,
218
+ "tool_switch_successful": 0
219
+ },
220
+ "runtime": {
221
+ "avg_steps": 1855.6,
222
+ "error_count": 0
223
+ }
224
+ }
225
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/native_shards/001/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/native_shards/001/summary.json ADDED
@@ -0,0 +1,212 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 100600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 10,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.4,
50
+ "success_rate_ci": {
51
+ "mean": 0.4,
52
+ "ci_low": 0.2,
53
+ "ci_high": 0.6
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.81
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 29.746666666666258
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.32,
64
+ "hard_violation_rate": 0.32,
65
+ "drop_rate": 0.32,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 0.0001714971702966901
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.44,
74
+ "at_end_success_rate": 0.44,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.4
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.8,
81
+ "current_chain_depth_progress_score": 0.7999999999999999,
82
+ "latched_chain_depth_progress_score": 0.8133333333333334,
83
+ "avg_current_chain_depth": 2.4,
84
+ "avg_latched_chain_depth": 2.44
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7180945684005536,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1784.8,
91
+ "avg_policy_steps_to_success": 595.2,
92
+ "avg_policy_queries_to_success": 595.2
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 0.0001714971702966901,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 7,
98
+ "drop_to_table_obj_013_apple_3": 1
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.4,
102
+ "joint_limit_active_step_rate": 0.09267278339907392,
103
+ "finger_joint_limit_saturation_step_rate": 0.0022723375064311437,
104
+ "max_joint_limit_excess_rad": 0.1570776253938675,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "left_thumb_2_joint": 4,
108
+ "right_little_1_joint": 798,
109
+ "right_middle_1_joint": 1,
110
+ "right_thumb_2_joint": 3871
111
+ },
112
+ "joint_limit_hard_counts": {
113
+ "left_ring_1_joint": 2,
114
+ "left_thumb_2_joint": 12,
115
+ "right_little_1_joint": 46,
116
+ "right_middle_1_joint": 2,
117
+ "right_thumb_2_joint": 311
118
+ },
119
+ "joint_limit_hard_episode_events": 332,
120
+ "finger_joint_limit_saturation_counts": {
121
+ "right_little_1_joint": 83,
122
+ "right_thumb_2_joint": 23
123
+ },
124
+ "joint_limit_max_excess_by_joint": {
125
+ "left_index_2_joint": 5.522568002902517e-08,
126
+ "left_little_1_joint": 0.0034786462783813477,
127
+ "left_little_2_joint": 2.452929770413448e-08,
128
+ "left_ring_1_joint": 0.006926059722900391,
129
+ "left_ring_2_joint": 1.3399794340784865e-07,
130
+ "left_thumb_1_joint": 0.002695918083190918,
131
+ "left_thumb_2_joint": 0.026098012924194336,
132
+ "left_thumb_3_joint": 0.0028136372566223145,
133
+ "left_thumb_4_joint": 0.0012449324131011963,
134
+ "right_index_1_joint": 0.002930909162387252,
135
+ "right_index_2_joint": 4.455142516235355e-06,
136
+ "right_little_1_joint": 0.1570776253938675,
137
+ "right_little_2_joint": 1.7450083760195412e-05,
138
+ "right_middle_1_joint": 0.01039359625428915,
139
+ "right_middle_2_joint": 4.99463612868567e-06,
140
+ "right_ring_1_joint": 0.0033123467583209276,
141
+ "right_ring_2_joint": 5.872071596968453e-06,
142
+ "right_thumb_1_joint": 0.0020017623901367188,
143
+ "right_thumb_2_joint": 0.11635972559452057,
144
+ "right_thumb_3_joint": 0.0042286585085093975
145
+ },
146
+ "coverage": {
147
+ "hard_limit_source": {
148
+ "available": 25,
149
+ "unavailable": 0,
150
+ "unknown": 0
151
+ },
152
+ "joint_role_manifest": {
153
+ "available": 25,
154
+ "unavailable": 0,
155
+ "unknown": 0
156
+ }
157
+ },
158
+ "hard_limit_sources": {
159
+ "runtime.joint_pos_limits": 25
160
+ },
161
+ "soft_limit_sources": {
162
+ "runtime.soft_joint_pos_limits": 25
163
+ },
164
+ "unknown_runtime_joint_names": [],
165
+ "baseline_checks": {
166
+ "post_home": {
167
+ "checked": 25,
168
+ "hard_violation_episodes": 0
169
+ },
170
+ "post_reset": {
171
+ "checked": 25,
172
+ "hard_violation_episodes": 0
173
+ }
174
+ }
175
+ }
176
+ },
177
+ "robot_motion": {
178
+ "joint_acceleration_rms": 31.660115977596554,
179
+ "joint_effort_rms": 28.85502941244965,
180
+ "joint_jerk_rms": 3248.2788953607424,
181
+ "joint_velocity_rms": 0.7983985710125173
182
+ },
183
+ "grasp": {
184
+ "kinematic_grasp_stability_index": 0.8196052258784088,
185
+ "mean_kinematic_grasp_stability": 0.23028360560495775,
186
+ "per_object_best_kinematic_grasp_stability": {
187
+ "obj_011_banana_2": 0.5120635795913414,
188
+ "obj_013_apple_3": 0.6150672858358699,
189
+ "obj_013_apple_4": 0.5354784130339514
190
+ },
191
+ "grasp_gsi_diagnostics": {
192
+ "mean_anti_slip_score": 0.8961749470717981,
193
+ "mean_duration_score": 0.22793971858196266,
194
+ "mean_hold_score": 0.09732398919614975,
195
+ "mean_kinematic_stability_score": 0.056484676269940075,
196
+ "mean_legacy_gsi": 0.29057969598789957,
197
+ "mean_lift_score": 0.19543717697076493,
198
+ "mean_motion_stability_score": 0.09867955586932954
199
+ }
200
+ },
201
+ "tool": {
202
+ "tool_selection_accuracy": null,
203
+ "tool_switch_success_rate": null,
204
+ "tool_switch_total": 0,
205
+ "tool_switch_successful": 0
206
+ },
207
+ "runtime": {
208
+ "avg_steps": 1865.92,
209
+ "error_count": 0
210
+ }
211
+ }
212
+ }
pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/06_fruit_bowl_loading/None/summary.json ADDED
@@ -0,0 +1,226 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA fruit_controller-429ffea065",
5
+ "base_seed": 100600000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 26,
10
+ "error_count": 0,
11
+ "evaluation_protocol": "reach_and_stop",
12
+ "early_stop": true,
13
+ "success_definition": "terminal.raw_condition holds for dwell_time_s",
14
+ "measurement_window": "episode_start_to_stable_success_or_failure_budget",
15
+ "max_episode_steps": 1920,
16
+ "dwell_time_s": 0.5,
17
+ "field_semantics": {
18
+ "success_rate": "primary protocol success rate: stable_success for reach_and_stop; terminal at_end_success_budget for fixed_horizon",
19
+ "avg_time_to_success_s": "main efficiency metric: mean time_to_stable_success_s over reached-success episodes",
20
+ "task_efficiency": "benchmark-specific expert-normalized speed ratio: expert_time_s / time_to_stable_success_s over reached-success episodes",
21
+ "avg_steps_to_success": "diagnostic mean steps_to_stable_success over reached-success episodes, or None if no episode succeeds",
22
+ "avg_policy_steps_to_success": "diagnostic mean policy-rate control/action steps through stable success over reached-success episodes",
23
+ "avg_policy_queries_to_success": "diagnostic mean actual model inference queries through stable success over reached-success episodes; None when the first-success query count was not observed",
24
+ "safe_success_rate": "episode success rate with no task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
25
+ "hard_violation_rate": "episode rate with any task-level hard safety violation (drop or tracked-object high speed; joint limits are diagnostic by default)",
26
+ "drop_rate": "episode rate with drop_off_table or drop_to_table_after_lift violation",
27
+ "high_speed_violation_rate": "episode rate with any tracked object exceeding the configured speed threshold",
28
+ "excessive_impact_rate": "deprecated contact-force impact metric; no longer populated by core safety",
29
+ "robot_constraint_violation_rate": "diagnostic episode rate with any asset hard-joint-limit observation; excluded from safe_success_rate unless include_in_core_safety is explicitly enabled",
30
+ "finger_joint_limit_saturation_rate": "episode rate with dexterous finger joints saturated beyond finger joint-limit thresholds; diagnostic by default and not included in safe_success_rate",
31
+ "joint_limit_active_step_rate": "diagnostic mean fraction of steps with any joint exceeding the soft diagnostic margin",
32
+ "finger_joint_limit_saturation_step_rate": "diagnostic mean fraction of steps with finger joint-limit saturation",
33
+ "max_joint_limit_excess_rad": "diagnostic maximum operating/soft-limit excess observed in radians; hard-limit source and role validation are in robot_constraint_diagnostics",
34
+ "safety_violation_rate_time": "compatibility diagnostic; with fixed physics_dt this is equivalent to safety_violation_step_rate and is omitted from grouped summary output",
35
+ "safety_violation_step_rate": "core safety metric: fraction of executed steps with task-level safety violations; joint-limit observations are excluded by default",
36
+ "safety_violation_rate_event": "deprecated compatibility alias of safety_violation_events_per_step; omitted from grouped summary output",
37
+ "safety_violation_events_per_step": "diagnostic event density; may exceed 1 when multiple violations happen in one step",
38
+ "episode_violation_rate": "compatibility alias of hard_violation_rate; omitted from grouped core output",
39
+ "latched_stage_completion_rate": "main progress metric; ever-reached stage completion rate at episode end",
40
+ "current_stage_completion_rate": "diagnostic terminal-state stage completion rate at episode end",
41
+ "current_normalized_progress_score": "chain-depth diagnostic; normalized progress score at episode end",
42
+ "current_chain_depth_progress_score": "alias of current_normalized_progress_score for ordered dependency-depth diagnostics",
43
+ "latched_chain_depth_progress_score": "alias of latched_normalized_progress_score for ever-reached dependency-depth diagnostics",
44
+ "robustness": "success rates grouped by benchmark protocol category: cov, inv, and inv+cov"
45
+ }
46
+ },
47
+ "core_metrics": {
48
+ "completion": {
49
+ "success_rate": 0.52,
50
+ "success_rate_ci": {
51
+ "mean": 0.52,
52
+ "ci_low": 0.38,
53
+ "ci_high": 0.66
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.87
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 30.101282051281622
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.46,
64
+ "hard_violation_rate": 0.24,
65
+ "drop_rate": 0.22,
66
+ "high_speed_violation_rate": 0.02,
67
+ "safety_violation_step_rate": 0.00012897955674025667
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.6,
74
+ "at_end_success_rate": 0.54,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.52
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.85,
81
+ "current_chain_depth_progress_score": 0.84,
82
+ "latched_chain_depth_progress_score": 0.8666666666666667,
83
+ "avg_current_chain_depth": 2.52,
84
+ "avg_latched_chain_depth": 2.6
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7094875095873755,
88
+ "expert_time_s": 21.333333333333332,
89
+ "expert_time_step": 1280,
90
+ "avg_steps_to_success": 1806.076923076923,
91
+ "avg_policy_steps_to_success": 602.3846153846154,
92
+ "avg_policy_queries_to_success": 602.3846153846154
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 0.00012897955674025667,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_011_banana_2": 10,
98
+ "drop_to_table_obj_013_apple_3": 1,
99
+ "high_speed_obj_024_bowl_1": 1
100
+ },
101
+ "robot_constraint_violation_rate": 1.0,
102
+ "finger_joint_limit_saturation_rate": 0.36,
103
+ "joint_limit_active_step_rate": 0.08731915991315377,
104
+ "finger_joint_limit_saturation_step_rate": 0.002375373503299727,
105
+ "max_joint_limit_excess_rad": 0.2602928578853607,
106
+ "robot_constraint_diagnostics": {
107
+ "joint_limit_soft_counts": {
108
+ "left_little_1_joint": 1,
109
+ "left_ring_1_joint": 71,
110
+ "left_thumb_2_joint": 139,
111
+ "right_index_1_joint": 62,
112
+ "right_little_1_joint": 1723,
113
+ "right_middle_1_joint": 67,
114
+ "right_ring_1_joint": 61,
115
+ "right_thumb_2_joint": 6706
116
+ },
117
+ "joint_limit_hard_counts": {
118
+ "left_little_1_joint": 4,
119
+ "left_ring_1_joint": 23,
120
+ "left_thumb_2_joint": 47,
121
+ "right_index_1_joint": 2,
122
+ "right_little_1_joint": 95,
123
+ "right_middle_1_joint": 7,
124
+ "right_ring_1_joint": 2,
125
+ "right_thumb_2_joint": 538,
126
+ "right_thumb_3_joint": 2
127
+ },
128
+ "joint_limit_hard_episode_events": 633,
129
+ "finger_joint_limit_saturation_counts": {
130
+ "left_ring_1_joint": 23,
131
+ "left_thumb_2_joint": 7,
132
+ "right_index_1_joint": 28,
133
+ "right_little_1_joint": 121,
134
+ "right_middle_1_joint": 31,
135
+ "right_ring_1_joint": 33,
136
+ "right_thumb_2_joint": 35
137
+ },
138
+ "joint_limit_max_excess_by_joint": {
139
+ "left_index_2_joint": 5.587289564346065e-08,
140
+ "left_little_1_joint": 0.011418342590332031,
141
+ "left_little_2_joint": 2.727411363423471e-08,
142
+ "left_ring_1_joint": 0.18063080310821533,
143
+ "left_ring_2_joint": 1.3556625333421835e-07,
144
+ "left_thumb_1_joint": 0.002695918083190918,
145
+ "left_thumb_2_joint": 0.11521251499652863,
146
+ "left_thumb_3_joint": 0.004622459411621094,
147
+ "left_thumb_4_joint": 0.001858144998550415,
148
+ "right_index_1_joint": 0.21233883500099182,
149
+ "right_index_2_joint": 5.520537797565339e-06,
150
+ "right_little_1_joint": 0.15717044472694397,
151
+ "right_little_2_joint": 1.7450083760195412e-05,
152
+ "right_middle_1_joint": 0.24825574457645416,
153
+ "right_middle_2_joint": 1.0853015737666283e-05,
154
+ "right_ring_1_joint": 0.2602928578853607,
155
+ "right_ring_2_joint": 1.1067111699958332e-05,
156
+ "right_thumb_1_joint": 0.0020017623901367188,
157
+ "right_thumb_2_joint": 0.11635972559452057,
158
+ "right_thumb_3_joint": 0.006209963466972113
159
+ },
160
+ "coverage": {
161
+ "hard_limit_source": {
162
+ "available": 50,
163
+ "unavailable": 0,
164
+ "unknown": 0
165
+ },
166
+ "joint_role_manifest": {
167
+ "available": 50,
168
+ "unavailable": 0,
169
+ "unknown": 0
170
+ }
171
+ },
172
+ "hard_limit_sources": {
173
+ "runtime.joint_pos_limits": 50
174
+ },
175
+ "soft_limit_sources": {
176
+ "runtime.soft_joint_pos_limits": 50
177
+ },
178
+ "unknown_runtime_joint_names": [],
179
+ "baseline_checks": {
180
+ "post_home": {
181
+ "checked": 50,
182
+ "hard_violation_episodes": 0
183
+ },
184
+ "post_reset": {
185
+ "checked": 50,
186
+ "hard_violation_episodes": 0
187
+ }
188
+ }
189
+ }
190
+ },
191
+ "robot_motion": {
192
+ "joint_acceleration_rms": 33.306819630704,
193
+ "joint_effort_rms": 29.499972576075887,
194
+ "joint_jerk_rms": 3412.4541117551075,
195
+ "joint_velocity_rms": 0.8290405721670079
196
+ },
197
+ "grasp": {
198
+ "kinematic_grasp_stability_index": 0.8500940525186795,
199
+ "mean_kinematic_grasp_stability": 0.23743974617718722,
200
+ "per_object_best_kinematic_grasp_stability": {
201
+ "obj_011_banana_2": 0.5815648543903286,
202
+ "obj_013_apple_3": 0.6840101556593694,
203
+ "obj_013_apple_4": 0.4476938710922983
204
+ },
205
+ "grasp_gsi_diagnostics": {
206
+ "mean_anti_slip_score": 0.8816973470304835,
207
+ "mean_duration_score": 0.22568086515369185,
208
+ "mean_hold_score": 0.11665752014255631,
209
+ "mean_kinematic_stability_score": 0.05863579379986997,
210
+ "mean_legacy_gsi": 0.2992892652094349,
211
+ "mean_lift_score": 0.22247162886626468,
212
+ "mean_motion_stability_score": 0.11708425183885637
213
+ }
214
+ },
215
+ "tool": {
216
+ "tool_selection_accuracy": null,
217
+ "tool_switch_success_rate": null,
218
+ "tool_switch_total": 0,
219
+ "tool_switch_successful": 0
220
+ },
221
+ "runtime": {
222
+ "avg_steps": 1860.76,
223
+ "error_count": 0
224
+ }
225
+ }
226
+ }