VME50121 commited on
Commit
79e4375
·
verified ·
1 Parent(s): 30409a0

Add PyRUA-Dex evaluation logs (part 3)

Browse files

Evaluation logs for the Bench2Dex leaderboard entry `pyrua-dex` (PKU DAGroup), a coding-agent system that uses no human demonstration data. GitHub PR: https://github.com/Bench2Dex/Bench2Dex-Leaderboard/pull/7

- `pyrua-dex/README.md`, `pyrua-dex/leaderboard_pyrua-dex.json` (identical to the GitHub results JSON)
- `pyrua-dex/per_task_eval/<task>/<None|Equi|Inv|Full>/per_episode.jsonl` and `summary.json`: the original Bench2Dex outputs of each channel's 50 episodes; `native_shards/000` and `001` hold the two batches of 25 as Bench2Dex wrote them.
- Sanitization: absolute asset paths are made relative (`dex2bench_dataset/...`); nothing else is changed (structure, fields, units, seeds, episode counts as written).
- Coverage: 26 tasks x 4 channels x 50 episodes = 5,200 episodes; Bench2Dex `ae870fb0e1f6237f20c8302198ee3850118892e1`. Overall SR 44.62, LSCR 67.02, Eff 0.8889 (Score 59.126).

🤖 Generated with [Claude Code](https://claude.com/claude-code)

This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. pyrua-dex/per_task_eval/64_sports_ball_cup_sort/None/native_shards/000/per_episode.jsonl +0 -0
  2. pyrua-dex/per_task_eval/64_sports_ball_cup_sort/None/native_shards/000/summary.json +195 -0
  3. pyrua-dex/per_task_eval/64_sports_ball_cup_sort/None/native_shards/001/per_episode.jsonl +0 -0
  4. pyrua-dex/per_task_eval/64_sports_ball_cup_sort/None/native_shards/001/summary.json +192 -0
  5. pyrua-dex/per_task_eval/64_sports_ball_cup_sort/None/per_episode.jsonl +0 -0
  6. pyrua-dex/per_task_eval/64_sports_ball_cup_sort/None/summary.json +196 -0
  7. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Equi/native_shards/000/per_episode.jsonl +0 -0
  8. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Equi/native_shards/000/summary.json +232 -0
  9. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Equi/native_shards/001/per_episode.jsonl +0 -0
  10. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Equi/native_shards/001/summary.json +238 -0
  11. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Equi/per_episode.jsonl +0 -0
  12. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Equi/summary.json +242 -0
  13. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Full/native_shards/000/per_episode.jsonl +0 -0
  14. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Full/native_shards/000/summary.json +226 -0
  15. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Full/native_shards/001/per_episode.jsonl +0 -0
  16. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Full/native_shards/001/summary.json +233 -0
  17. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Full/per_episode.jsonl +0 -0
  18. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Full/summary.json +238 -0
  19. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Inv/native_shards/000/per_episode.jsonl +0 -0
  20. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Inv/native_shards/000/summary.json +231 -0
  21. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Inv/native_shards/001/per_episode.jsonl +0 -0
  22. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Inv/native_shards/001/summary.json +236 -0
  23. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Inv/per_episode.jsonl +0 -0
  24. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Inv/summary.json +237 -0
  25. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/None/native_shards/000/per_episode.jsonl +0 -0
  26. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/None/native_shards/000/summary.json +236 -0
  27. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/None/native_shards/001/per_episode.jsonl +0 -0
  28. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/None/native_shards/001/summary.json +234 -0
  29. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/None/per_episode.jsonl +0 -0
  30. pyrua-dex/per_task_eval/67_faucet_cup_water_fill/None/summary.json +241 -0
  31. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Equi/native_shards/000/per_episode.jsonl +0 -0
  32. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Equi/native_shards/000/summary.json +220 -0
  33. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Equi/native_shards/001/per_episode.jsonl +0 -0
  34. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Equi/native_shards/001/summary.json +218 -0
  35. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Equi/per_episode.jsonl +0 -0
  36. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Equi/summary.json +221 -0
  37. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Full/native_shards/000/per_episode.jsonl +0 -0
  38. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Full/native_shards/000/summary.json +217 -0
  39. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Full/native_shards/001/per_episode.jsonl +0 -0
  40. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Full/native_shards/001/summary.json +218 -0
  41. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Full/per_episode.jsonl +0 -0
  42. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Full/summary.json +218 -0
  43. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Inv/native_shards/000/per_episode.jsonl +0 -0
  44. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Inv/native_shards/000/summary.json +217 -0
  45. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Inv/native_shards/001/per_episode.jsonl +0 -0
  46. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Inv/native_shards/001/summary.json +218 -0
  47. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Inv/per_episode.jsonl +0 -0
  48. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Inv/summary.json +218 -0
  49. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/None/native_shards/000/per_episode.jsonl +0 -0
  50. pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/None/native_shards/000/summary.json +222 -0
pyrua-dex/per_task_eval/64_sports_ball_cup_sort/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/64_sports_ball_cup_sort/None/native_shards/000/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 ball_sort-7a836c8a10",
5
+ "base_seed": 106400000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 8,
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": 1995,
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": 0.32,
50
+ "success_rate_ci": {
51
+ "mean": 0.32,
52
+ "ci_low": 0.16,
53
+ "ci_high": 0.52
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.44
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 28.07291666666635
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.32,
64
+ "hard_violation_rate": 0.16,
65
+ "drop_rate": 0.16,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 0.00012660898923823593
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.32,
74
+ "at_end_success_rate": 0.32,
75
+ "task_family_success_rate": {
76
+ "rigid_pick_place": 0.32
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.42,
81
+ "current_chain_depth_progress_score": 0.42,
82
+ "latched_chain_depth_progress_score": 0.44,
83
+ "avg_current_chain_depth": 1.68,
84
+ "avg_latched_chain_depth": 1.76
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7895892411251773,
88
+ "expert_time_s": 22.15,
89
+ "expert_time_step": 1329,
90
+ "avg_steps_to_success": 1684.375,
91
+ "avg_policy_steps_to_success": 561.75,
92
+ "avg_policy_queries_to_success": 561.75
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 0.00012660898923823593,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_054_softball_3": 3,
98
+ "drop_to_table_obj_055_baseball_4": 2,
99
+ "drop_to_table_obj_057_racquetball_5": 1
100
+ },
101
+ "robot_constraint_violation_rate": 0.2,
102
+ "finger_joint_limit_saturation_rate": 0.04,
103
+ "joint_limit_active_step_rate": 0.0027009917704156994,
104
+ "finger_joint_limit_saturation_step_rate": 0.0001688119856509812,
105
+ "max_joint_limit_excess_rad": 0.14719152450561523,
106
+ "robot_constraint_diagnostics": {
107
+ "joint_limit_soft_counts": {
108
+ "multi_joint_10_0": 96,
109
+ "multi_joint_9_0": 48
110
+ },
111
+ "joint_limit_hard_counts": {
112
+ "multi_joint_10_0": 6,
113
+ "multi_joint_9_0": 2
114
+ },
115
+ "joint_limit_hard_episode_events": 6,
116
+ "finger_joint_limit_saturation_counts": {
117
+ "multi_joint_10_0": 8
118
+ },
119
+ "joint_limit_max_excess_by_joint": {
120
+ "joint_12_0": 0.002146005630493164,
121
+ "joint_6_0": 0.0028853416442871094,
122
+ "joint_9_0": 2.1457672119140625e-05,
123
+ "multi_joint_10_0": 0.14719152450561523,
124
+ "multi_joint_12_0": 0.0008533000946044922,
125
+ "multi_joint_2_0": 0.0037798285484313965,
126
+ "multi_joint_9_0": 0.09094870090484619
127
+ },
128
+ "coverage": {
129
+ "hard_limit_source": {
130
+ "available": 25,
131
+ "unavailable": 0,
132
+ "unknown": 0
133
+ },
134
+ "joint_role_manifest": {
135
+ "available": 25,
136
+ "unavailable": 0,
137
+ "unknown": 0
138
+ }
139
+ },
140
+ "hard_limit_sources": {
141
+ "runtime.joint_pos_limits": 25
142
+ },
143
+ "soft_limit_sources": {
144
+ "runtime.soft_joint_pos_limits": 25
145
+ },
146
+ "unknown_runtime_joint_names": [],
147
+ "baseline_checks": {
148
+ "post_home": {
149
+ "checked": 25,
150
+ "hard_violation_episodes": 0
151
+ },
152
+ "post_reset": {
153
+ "checked": 25,
154
+ "hard_violation_episodes": 0
155
+ }
156
+ }
157
+ }
158
+ },
159
+ "robot_motion": {
160
+ "joint_acceleration_rms": 4.935946719979667,
161
+ "joint_effort_rms": 16.801369717282384,
162
+ "joint_jerk_rms": 501.45790328886017,
163
+ "joint_velocity_rms": 0.15531522180678908
164
+ },
165
+ "grasp": {
166
+ "kinematic_grasp_stability_index": 0.5303771293595924,
167
+ "mean_kinematic_grasp_stability": 0.25665202746692506,
168
+ "per_object_best_kinematic_grasp_stability": {
169
+ "obj_054_softball_3": 0.43385713730750075,
170
+ "obj_055_baseball_4": 0.29626944663393395,
171
+ "obj_057_racquetball_5": 0.14936474011613984,
172
+ "obj_058_golf_ball_6": 0.22493906420260695
173
+ },
174
+ "grasp_gsi_diagnostics": {
175
+ "mean_anti_slip_score": 0.735261102203261,
176
+ "mean_duration_score": 0.3085796167955244,
177
+ "mean_hold_score": 0.27640121839695175,
178
+ "mean_kinematic_stability_score": 0.04331032827469376,
179
+ "mean_legacy_gsi": 0.3649031740449593,
180
+ "mean_lift_score": 0.32834005663258364,
181
+ "mean_motion_stability_score": 0.24911202358061457
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": 1895.6,
192
+ "error_count": 0
193
+ }
194
+ }
195
+ }
pyrua-dex/per_task_eval/64_sports_ball_cup_sort/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/64_sports_ball_cup_sort/None/native_shards/001/summary.json ADDED
@@ -0,0 +1,192 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA ball_sort-7a836c8a10",
5
+ "base_seed": 106400000,
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": 1995,
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": 0.63
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 27.928205128204823
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.52,
64
+ "hard_violation_rate": 0.04,
65
+ "drop_rate": 0.04,
66
+ "high_speed_violation_rate": 0.04,
67
+ "safety_violation_step_rate": 0.00013122211530049865
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.52,
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": 0.61,
81
+ "current_chain_depth_progress_score": 0.63,
82
+ "latched_chain_depth_progress_score": 0.63,
83
+ "avg_current_chain_depth": 2.52,
84
+ "avg_latched_chain_depth": 2.52
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.793376252679206,
88
+ "expert_time_s": 22.15,
89
+ "expert_time_step": 1329,
90
+ "avg_steps_to_success": 1675.6923076923076,
91
+ "avg_policy_steps_to_success": 558.7692307692307,
92
+ "avg_policy_queries_to_success": 558.7692307692307
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 6.561105765024933e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_054_softball_3": 1,
98
+ "drop_to_table_obj_055_baseball_4": 1,
99
+ "high_speed_obj_055_baseball_4": 1
100
+ },
101
+ "robot_constraint_violation_rate": 0.44,
102
+ "finger_joint_limit_saturation_rate": 0.0,
103
+ "joint_limit_active_step_rate": 0.0038054413437144607,
104
+ "finger_joint_limit_saturation_step_rate": 0.0,
105
+ "max_joint_limit_excess_rad": 0.15495535731315613,
106
+ "robot_constraint_diagnostics": {
107
+ "joint_limit_soft_counts": {
108
+ "multi_joint_10_0": 109,
109
+ "multi_joint_9_0": 110
110
+ },
111
+ "joint_limit_hard_counts": {
112
+ "multi_joint_10_0": 18,
113
+ "multi_joint_9_0": 9
114
+ },
115
+ "joint_limit_hard_episode_events": 18,
116
+ "finger_joint_limit_saturation_counts": {},
117
+ "joint_limit_max_excess_by_joint": {
118
+ "joint_12_0": 0.0004987716674804688,
119
+ "joint_6_0": 0.0033805370330810547,
120
+ "multi_joint_10_0": 0.15495535731315613,
121
+ "multi_joint_12_0": 0.0010230541229248047,
122
+ "multi_joint_2_0": 0.0037820935249328613,
123
+ "multi_joint_9_0": 0.0722501277923584
124
+ },
125
+ "coverage": {
126
+ "hard_limit_source": {
127
+ "available": 25,
128
+ "unavailable": 0,
129
+ "unknown": 0
130
+ },
131
+ "joint_role_manifest": {
132
+ "available": 25,
133
+ "unavailable": 0,
134
+ "unknown": 0
135
+ }
136
+ },
137
+ "hard_limit_sources": {
138
+ "runtime.joint_pos_limits": 25
139
+ },
140
+ "soft_limit_sources": {
141
+ "runtime.soft_joint_pos_limits": 25
142
+ },
143
+ "unknown_runtime_joint_names": [],
144
+ "baseline_checks": {
145
+ "post_home": {
146
+ "checked": 25,
147
+ "hard_violation_episodes": 0
148
+ },
149
+ "post_reset": {
150
+ "checked": 25,
151
+ "hard_violation_episodes": 0
152
+ }
153
+ }
154
+ }
155
+ },
156
+ "robot_motion": {
157
+ "joint_acceleration_rms": 5.337204213604483,
158
+ "joint_effort_rms": 17.373254420261915,
159
+ "joint_jerk_rms": 538.3611620187448,
160
+ "joint_velocity_rms": 0.17625382065104267
161
+ },
162
+ "grasp": {
163
+ "kinematic_grasp_stability_index": 0.6279293032993509,
164
+ "mean_kinematic_grasp_stability": 0.30333262198395794,
165
+ "per_object_best_kinematic_grasp_stability": {
166
+ "obj_054_softball_3": 0.4962688400159625,
167
+ "obj_055_baseball_4": 0.325518320003637,
168
+ "obj_057_racquetball_5": 0.24858750991898665,
169
+ "obj_058_golf_ball_6": 0.3247033422849403
170
+ },
171
+ "grasp_gsi_diagnostics": {
172
+ "mean_anti_slip_score": 0.6953697975667725,
173
+ "mean_duration_score": 0.3549257611460195,
174
+ "mean_hold_score": 0.32415515445727133,
175
+ "mean_kinematic_stability_score": 0.050726727048377,
176
+ "mean_legacy_gsi": 0.39271714199341384,
177
+ "mean_lift_score": 0.38234688979834147,
178
+ "mean_motion_stability_score": 0.2905667260829128
179
+ }
180
+ },
181
+ "tool": {
182
+ "tool_selection_accuracy": null,
183
+ "tool_switch_success_rate": null,
184
+ "tool_switch_total": 0,
185
+ "tool_switch_successful": 0
186
+ },
187
+ "runtime": {
188
+ "avg_steps": 1828.96,
189
+ "error_count": 0
190
+ }
191
+ }
192
+ }
pyrua-dex/per_task_eval/64_sports_ball_cup_sort/None/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/64_sports_ball_cup_sort/None/summary.json ADDED
@@ -0,0 +1,196 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA ball_sort-7a836c8a10",
5
+ "base_seed": 106400000,
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": 1995,
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": 0.535
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 27.983333333333025
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.42,
64
+ "hard_violation_rate": 0.1,
65
+ "drop_rate": 0.1,
66
+ "high_speed_violation_rate": 0.02,
67
+ "safety_violation_step_rate": 0.00012887428313680004
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.42,
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": 0.515,
81
+ "current_chain_depth_progress_score": 0.525,
82
+ "latched_chain_depth_progress_score": 0.535,
83
+ "avg_current_chain_depth": 2.1,
84
+ "avg_latched_chain_depth": 2.14
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7919335816110046,
88
+ "expert_time_s": 22.15,
89
+ "expert_time_step": 1329,
90
+ "avg_steps_to_success": 1679,
91
+ "avg_policy_steps_to_success": 559.9047619047619,
92
+ "avg_policy_queries_to_success": 559.9047619047619
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 9.665571235260003e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_054_softball_3": 4,
98
+ "drop_to_table_obj_055_baseball_4": 3,
99
+ "drop_to_table_obj_057_racquetball_5": 1,
100
+ "high_speed_obj_055_baseball_4": 1
101
+ },
102
+ "robot_constraint_violation_rate": 0.32,
103
+ "finger_joint_limit_saturation_rate": 0.02,
104
+ "joint_limit_active_step_rate": 0.003243336125609468,
105
+ "finger_joint_limit_saturation_step_rate": 8.59161887578667e-05,
106
+ "max_joint_limit_excess_rad": 0.15495535731315613,
107
+ "robot_constraint_diagnostics": {
108
+ "joint_limit_soft_counts": {
109
+ "multi_joint_10_0": 205,
110
+ "multi_joint_9_0": 158
111
+ },
112
+ "joint_limit_hard_counts": {
113
+ "multi_joint_10_0": 24,
114
+ "multi_joint_9_0": 11
115
+ },
116
+ "joint_limit_hard_episode_events": 24,
117
+ "finger_joint_limit_saturation_counts": {
118
+ "multi_joint_10_0": 8
119
+ },
120
+ "joint_limit_max_excess_by_joint": {
121
+ "joint_12_0": 0.002146005630493164,
122
+ "joint_6_0": 0.0033805370330810547,
123
+ "joint_9_0": 2.1457672119140625e-05,
124
+ "multi_joint_10_0": 0.15495535731315613,
125
+ "multi_joint_12_0": 0.0010230541229248047,
126
+ "multi_joint_2_0": 0.0037820935249328613,
127
+ "multi_joint_9_0": 0.09094870090484619
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": 5.136575466792076,
162
+ "joint_effort_rms": 17.08731206877215,
163
+ "joint_jerk_rms": 519.9095326538026,
164
+ "joint_velocity_rms": 0.16578452122891588
165
+ },
166
+ "grasp": {
167
+ "kinematic_grasp_stability_index": 0.5791532163294717,
168
+ "mean_kinematic_grasp_stability": 0.2810071202584205,
169
+ "per_object_best_kinematic_grasp_stability": {
170
+ "obj_054_softball_3": 0.4650629886617316,
171
+ "obj_055_baseball_4": 0.3108938833187855,
172
+ "obj_057_racquetball_5": 0.19897612501756323,
173
+ "obj_058_golf_ball_6": 0.2748212032437736
174
+ },
175
+ "grasp_gsi_diagnostics": {
176
+ "mean_anti_slip_score": 0.7153154498850168,
177
+ "mean_duration_score": 0.331752688970772,
178
+ "mean_hold_score": 0.30027818642711157,
179
+ "mean_kinematic_stability_score": 0.047018527661535386,
180
+ "mean_legacy_gsi": 0.3788101580191866,
181
+ "mean_lift_score": 0.35534347321546256,
182
+ "mean_motion_stability_score": 0.2698393748317637
183
+ }
184
+ },
185
+ "tool": {
186
+ "tool_selection_accuracy": null,
187
+ "tool_switch_success_rate": null,
188
+ "tool_switch_total": 0,
189
+ "tool_switch_successful": 0
190
+ },
191
+ "runtime": {
192
+ "avg_steps": 1862.28,
193
+ "error_count": 0
194
+ }
195
+ }
196
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/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/67_faucet_cup_water_fill/Equi/native_shards/000/summary.json ADDED
@@ -0,0 +1,232 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA faucet_recovery-ee8f8b886b",
5
+ "base_seed": 106700000,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.84
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 41.0731481481471
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.68,
64
+ "hard_violation_rate": 0.08,
65
+ "drop_rate": 0.08,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 3.0958314629351575e-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
+ "articulation_multi": 0.72
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.63,
81
+ "current_chain_depth_progress_score": 0.83,
82
+ "latched_chain_depth_progress_score": 0.84,
83
+ "avg_current_chain_depth": 3.32,
84
+ "avg_latched_chain_depth": 3.36
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7843610324509178,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2464.3888888888887,
91
+ "avg_policy_steps_to_success": 821.8888888888889,
92
+ "avg_policy_queries_to_success": 821.8888888888889
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 3.0958314629351575e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_025_mug_1": 1,
98
+ "drop_to_table_obj_031_spoon_1": 1
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.6,
102
+ "joint_limit_active_step_rate": 0.027444545918920173,
103
+ "finger_joint_limit_saturation_step_rate": 0.007368078881785676,
104
+ "max_joint_limit_excess_rad": 0.8790982961654663,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "l_joint4": 11,
108
+ "left_index_proximal_joint": 987,
109
+ "left_middle_distal_joint": 37,
110
+ "left_middle_proximal_joint": 30,
111
+ "left_pinky_distal_joint": 16,
112
+ "left_pinky_proximal_joint": 41,
113
+ "left_ring_distal_joint": 39,
114
+ "left_ring_proximal_joint": 70,
115
+ "left_thumb_distal_joint": 62,
116
+ "left_thumb_proximal_joint": 8,
117
+ "right_pinky_proximal_joint": 124,
118
+ "right_thumb_proximal_joint": 449
119
+ },
120
+ "joint_limit_hard_counts": {
121
+ "l_joint4": 1,
122
+ "left_index_proximal_joint": 201,
123
+ "left_middle_distal_joint": 47,
124
+ "left_middle_proximal_joint": 37,
125
+ "left_pinky_distal_joint": 44,
126
+ "left_pinky_proximal_joint": 184,
127
+ "left_ring_distal_joint": 59,
128
+ "left_ring_proximal_joint": 80,
129
+ "left_thumb_distal_joint": 5,
130
+ "left_thumb_proximal_joint": 1,
131
+ "right_middle_distal_joint": 1,
132
+ "right_middle_proximal_joint": 2,
133
+ "right_pinky_proximal_joint": 21,
134
+ "right_thumb_proximal_joint": 5
135
+ },
136
+ "joint_limit_hard_episode_events": 512,
137
+ "finger_joint_limit_saturation_counts": {
138
+ "left_index_proximal_joint": 374,
139
+ "left_pinky_proximal_joint": 2,
140
+ "left_ring_proximal_joint": 9,
141
+ "left_thumb_distal_joint": 16,
142
+ "right_pinky_proximal_joint": 86
143
+ },
144
+ "joint_limit_max_excess_by_joint": {
145
+ "l_joint4": 0.1283411979675293,
146
+ "left_index_distal_joint": 1.4893463351484115e-08,
147
+ "left_index_proximal_joint": 0.8790982961654663,
148
+ "left_middle_distal_joint": 0.11421144008636475,
149
+ "left_middle_proximal_joint": 0.16016316413879395,
150
+ "left_pinky_distal_joint": 0.0457988977432251,
151
+ "left_pinky_proximal_joint": 0.2703211307525635,
152
+ "left_ring_distal_joint": 0.06635820865631104,
153
+ "left_ring_proximal_joint": 0.2874271869659424,
154
+ "left_thumb_distal_joint": 0.6531891822814941,
155
+ "left_thumb_metacarpal_joint": 0.00014285073848441243,
156
+ "left_thumb_proximal_joint": 0.1261489987373352,
157
+ "right_index_distal_joint": 2.8387556194608976e-10,
158
+ "right_index_proximal_joint": 0.0015238194027915597,
159
+ "right_middle_distal_joint": 0.005158443935215473,
160
+ "right_middle_proximal_joint": 0.006842489819973707,
161
+ "right_pinky_distal_joint": 0.002419952768832445,
162
+ "right_pinky_proximal_joint": 0.4847389757633209,
163
+ "right_ring_distal_joint": 0.004035755060613155,
164
+ "right_ring_proximal_joint": 0.003577441442757845,
165
+ "right_thumb_proximal_joint": 0.061630845069885254
166
+ },
167
+ "coverage": {
168
+ "hard_limit_source": {
169
+ "available": 25,
170
+ "unavailable": 0,
171
+ "unknown": 0
172
+ },
173
+ "joint_role_manifest": {
174
+ "available": 25,
175
+ "unavailable": 0,
176
+ "unknown": 0
177
+ }
178
+ },
179
+ "hard_limit_sources": {
180
+ "runtime.joint_pos_limits": 25
181
+ },
182
+ "soft_limit_sources": {
183
+ "runtime.soft_joint_pos_limits": 25
184
+ },
185
+ "unknown_runtime_joint_names": [],
186
+ "baseline_checks": {
187
+ "post_home": {
188
+ "checked": 25,
189
+ "hard_violation_episodes": 0
190
+ },
191
+ "post_reset": {
192
+ "checked": 25,
193
+ "hard_violation_episodes": 0
194
+ }
195
+ }
196
+ }
197
+ },
198
+ "robot_motion": {
199
+ "joint_acceleration_rms": 15.27850633361912,
200
+ "joint_effort_rms": 26.190558852083985,
201
+ "joint_jerk_rms": 1563.8340974193713,
202
+ "joint_velocity_rms": 0.3626804318443024
203
+ },
204
+ "grasp": {
205
+ "kinematic_grasp_stability_index": 0.8651644137499624,
206
+ "mean_kinematic_grasp_stability": 0.3447765824080762,
207
+ "per_object_best_kinematic_grasp_stability": {
208
+ "obj_025_mug_1": 0.7910548727308129,
209
+ "obj_031_spoon_1": 0.7777271508721754
210
+ },
211
+ "grasp_gsi_diagnostics": {
212
+ "mean_anti_slip_score": 0.633148833171132,
213
+ "mean_duration_score": 0.6708367260925681,
214
+ "mean_hold_score": 0.4035720647900083,
215
+ "mean_kinematic_stability_score": 0.09817541484882884,
216
+ "mean_legacy_gsi": 0.4972890395537587,
217
+ "mean_lift_score": 0.5488246323819629,
218
+ "mean_motion_stability_score": 0.37712982582740934
219
+ }
220
+ },
221
+ "tool": {
222
+ "tool_selection_accuracy": null,
223
+ "tool_switch_success_rate": null,
224
+ "tool_switch_total": 0,
225
+ "tool_switch_successful": 0
226
+ },
227
+ "runtime": {
228
+ "avg_steps": 2584.12,
229
+ "error_count": 0
230
+ }
231
+ }
232
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/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/67_faucet_cup_water_fill/Equi/native_shards/001/summary.json ADDED
@@ -0,0 +1,238 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA faucet_recovery-ee8f8b886b",
5
+ "base_seed": 106700000,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.84
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 40.91851851851748
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": 3.1038549878949655e-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
+ "articulation_multi": 0.72
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.6,
81
+ "current_chain_depth_progress_score": 0.83,
82
+ "latched_chain_depth_progress_score": 0.84,
83
+ "avg_current_chain_depth": 3.32,
84
+ "avg_latched_chain_depth": 3.36
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7887397952961676,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2455.1111111111113,
91
+ "avg_policy_steps_to_success": 818.7777777777778,
92
+ "avg_policy_queries_to_success": 818.7777777777778
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 3.1038549878949655e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_031_spoon_1": 2
98
+ },
99
+ "robot_constraint_violation_rate": 1.0,
100
+ "finger_joint_limit_saturation_rate": 0.72,
101
+ "joint_limit_active_step_rate": 0.049413371407287854,
102
+ "finger_joint_limit_saturation_step_rate": 0.034639021664907815,
103
+ "max_joint_limit_excess_rad": 1.1781593561172485,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {
106
+ "left_index_proximal_joint": 653,
107
+ "left_middle_distal_joint": 101,
108
+ "left_middle_proximal_joint": 67,
109
+ "left_pinky_distal_joint": 173,
110
+ "left_pinky_proximal_joint": 193,
111
+ "left_ring_distal_joint": 193,
112
+ "left_ring_proximal_joint": 235,
113
+ "left_thumb_distal_joint": 224,
114
+ "left_thumb_metacarpal_joint": 200,
115
+ "left_thumb_proximal_joint": 10,
116
+ "right_middle_proximal_joint": 1,
117
+ "right_pinky_proximal_joint": 1753,
118
+ "right_thumb_proximal_joint": 29
119
+ },
120
+ "joint_limit_hard_counts": {
121
+ "left_index_proximal_joint": 161,
122
+ "left_middle_distal_joint": 76,
123
+ "left_middle_proximal_joint": 33,
124
+ "left_pinky_distal_joint": 79,
125
+ "left_pinky_proximal_joint": 211,
126
+ "left_ring_distal_joint": 106,
127
+ "left_ring_proximal_joint": 120,
128
+ "left_thumb_distal_joint": 13,
129
+ "left_thumb_metacarpal_joint": 4,
130
+ "left_thumb_proximal_joint": 1,
131
+ "right_middle_distal_joint": 2,
132
+ "right_middle_proximal_joint": 6,
133
+ "right_pinky_proximal_joint": 25,
134
+ "right_ring_distal_joint": 1,
135
+ "right_thumb_proximal_joint": 1
136
+ },
137
+ "joint_limit_hard_episode_events": 529,
138
+ "finger_joint_limit_saturation_counts": {
139
+ "left_index_proximal_joint": 247,
140
+ "left_middle_distal_joint": 9,
141
+ "left_middle_proximal_joint": 19,
142
+ "left_pinky_distal_joint": 92,
143
+ "left_pinky_proximal_joint": 9,
144
+ "left_ring_distal_joint": 35,
145
+ "left_ring_proximal_joint": 59,
146
+ "left_thumb_distal_joint": 97,
147
+ "left_thumb_metacarpal_joint": 182,
148
+ "right_pinky_proximal_joint": 1699,
149
+ "right_thumb_proximal_joint": 3
150
+ },
151
+ "joint_limit_max_excess_by_joint": {
152
+ "left_index_distal_joint": 1.008583438988353e-08,
153
+ "left_index_proximal_joint": 0.7296307682991028,
154
+ "left_middle_distal_joint": 0.4184095859527588,
155
+ "left_middle_proximal_joint": 0.3883697986602783,
156
+ "left_pinky_distal_joint": 0.24773669242858887,
157
+ "left_pinky_proximal_joint": 0.32277607917785645,
158
+ "left_ring_distal_joint": 0.20986711978912354,
159
+ "left_ring_proximal_joint": 0.5477006435394287,
160
+ "left_thumb_distal_joint": 1.0694735050201416,
161
+ "left_thumb_metacarpal_joint": 1.1781593561172485,
162
+ "left_thumb_proximal_joint": 0.13173270225524902,
163
+ "right_index_distal_joint": 2.8117541628347453e-10,
164
+ "right_index_proximal_joint": 0.0015330095775425434,
165
+ "right_middle_distal_joint": 0.007450763136148453,
166
+ "right_middle_proximal_joint": 0.011103007942438126,
167
+ "right_pinky_distal_joint": 0.0035635153762996197,
168
+ "right_pinky_proximal_joint": 0.5900799632072449,
169
+ "right_ring_distal_joint": 0.005387734156101942,
170
+ "right_ring_proximal_joint": 0.0033602656330913305,
171
+ "right_thumb_proximal_joint": 0.15371644496917725
172
+ },
173
+ "coverage": {
174
+ "hard_limit_source": {
175
+ "available": 25,
176
+ "unavailable": 0,
177
+ "unknown": 0
178
+ },
179
+ "joint_role_manifest": {
180
+ "available": 25,
181
+ "unavailable": 0,
182
+ "unknown": 0
183
+ }
184
+ },
185
+ "hard_limit_sources": {
186
+ "runtime.joint_pos_limits": 25
187
+ },
188
+ "soft_limit_sources": {
189
+ "runtime.soft_joint_pos_limits": 25
190
+ },
191
+ "unknown_runtime_joint_names": [],
192
+ "baseline_checks": {
193
+ "post_home": {
194
+ "checked": 25,
195
+ "hard_violation_episodes": 0
196
+ },
197
+ "post_reset": {
198
+ "checked": 25,
199
+ "hard_violation_episodes": 0
200
+ }
201
+ }
202
+ }
203
+ },
204
+ "robot_motion": {
205
+ "joint_acceleration_rms": 15.665273719283544,
206
+ "joint_effort_rms": 25.092101041572622,
207
+ "joint_jerk_rms": 1596.137647879776,
208
+ "joint_velocity_rms": 0.37090708613269974
209
+ },
210
+ "grasp": {
211
+ "kinematic_grasp_stability_index": 0.9277775116446052,
212
+ "mean_kinematic_grasp_stability": 0.36499151973121685,
213
+ "per_object_best_kinematic_grasp_stability": {
214
+ "obj_025_mug_1": 0.8430854596199872,
215
+ "obj_031_spoon_1": 0.8098030107286133
216
+ },
217
+ "grasp_gsi_diagnostics": {
218
+ "mean_anti_slip_score": 0.6590500939497907,
219
+ "mean_duration_score": 0.6947759068877236,
220
+ "mean_hold_score": 0.40670626670515575,
221
+ "mean_kinematic_stability_score": 0.12203296149625269,
222
+ "mean_legacy_gsi": 0.5077287476911215,
223
+ "mean_lift_score": 0.5472781249865064,
224
+ "mean_motion_stability_score": 0.3753372290634238
225
+ }
226
+ },
227
+ "tool": {
228
+ "tool_selection_accuracy": null,
229
+ "tool_switch_success_rate": null,
230
+ "tool_switch_total": 0,
231
+ "tool_switch_successful": 0
232
+ },
233
+ "runtime": {
234
+ "avg_steps": 2577.44,
235
+ "error_count": 0
236
+ }
237
+ }
238
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Equi/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Equi/summary.json ADDED
@@ -0,0 +1,242 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA faucet_recovery-ee8f8b886b",
5
+ "base_seed": 106700000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 36,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.6,
53
+ "ci_high": 0.84
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.84
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 40.995833333332286
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.7,
64
+ "hard_violation_rate": 0.08,
65
+ "drop_rate": 0.08,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 3.099838033462752e-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
+ "articulation_multi": 0.72
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.615,
81
+ "current_chain_depth_progress_score": 0.83,
82
+ "latched_chain_depth_progress_score": 0.84,
83
+ "avg_current_chain_depth": 3.32,
84
+ "avg_latched_chain_depth": 3.36
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7865504138735427,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2459.75,
91
+ "avg_policy_steps_to_success": 820.3333333333334,
92
+ "avg_policy_queries_to_success": 820.3333333333334
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 3.099838033462752e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_025_mug_1": 1,
98
+ "drop_to_table_obj_031_spoon_1": 3
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.66,
102
+ "joint_limit_active_step_rate": 0.03841474282968715,
103
+ "finger_joint_limit_saturation_step_rate": 0.020985903486542827,
104
+ "max_joint_limit_excess_rad": 1.1781593561172485,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "l_joint4": 11,
108
+ "left_index_proximal_joint": 1640,
109
+ "left_middle_distal_joint": 138,
110
+ "left_middle_proximal_joint": 97,
111
+ "left_pinky_distal_joint": 189,
112
+ "left_pinky_proximal_joint": 234,
113
+ "left_ring_distal_joint": 232,
114
+ "left_ring_proximal_joint": 305,
115
+ "left_thumb_distal_joint": 286,
116
+ "left_thumb_metacarpal_joint": 200,
117
+ "left_thumb_proximal_joint": 18,
118
+ "right_middle_proximal_joint": 1,
119
+ "right_pinky_proximal_joint": 1877,
120
+ "right_thumb_proximal_joint": 478
121
+ },
122
+ "joint_limit_hard_counts": {
123
+ "l_joint4": 1,
124
+ "left_index_proximal_joint": 362,
125
+ "left_middle_distal_joint": 123,
126
+ "left_middle_proximal_joint": 70,
127
+ "left_pinky_distal_joint": 123,
128
+ "left_pinky_proximal_joint": 395,
129
+ "left_ring_distal_joint": 165,
130
+ "left_ring_proximal_joint": 200,
131
+ "left_thumb_distal_joint": 18,
132
+ "left_thumb_metacarpal_joint": 4,
133
+ "left_thumb_proximal_joint": 2,
134
+ "right_middle_distal_joint": 3,
135
+ "right_middle_proximal_joint": 8,
136
+ "right_pinky_proximal_joint": 46,
137
+ "right_ring_distal_joint": 1,
138
+ "right_thumb_proximal_joint": 6
139
+ },
140
+ "joint_limit_hard_episode_events": 1041,
141
+ "finger_joint_limit_saturation_counts": {
142
+ "left_index_proximal_joint": 621,
143
+ "left_middle_distal_joint": 9,
144
+ "left_middle_proximal_joint": 19,
145
+ "left_pinky_distal_joint": 92,
146
+ "left_pinky_proximal_joint": 11,
147
+ "left_ring_distal_joint": 35,
148
+ "left_ring_proximal_joint": 68,
149
+ "left_thumb_distal_joint": 113,
150
+ "left_thumb_metacarpal_joint": 182,
151
+ "right_pinky_proximal_joint": 1785,
152
+ "right_thumb_proximal_joint": 3
153
+ },
154
+ "joint_limit_max_excess_by_joint": {
155
+ "l_joint4": 0.1283411979675293,
156
+ "left_index_distal_joint": 1.4893463351484115e-08,
157
+ "left_index_proximal_joint": 0.8790982961654663,
158
+ "left_middle_distal_joint": 0.4184095859527588,
159
+ "left_middle_proximal_joint": 0.3883697986602783,
160
+ "left_pinky_distal_joint": 0.24773669242858887,
161
+ "left_pinky_proximal_joint": 0.32277607917785645,
162
+ "left_ring_distal_joint": 0.20986711978912354,
163
+ "left_ring_proximal_joint": 0.5477006435394287,
164
+ "left_thumb_distal_joint": 1.0694735050201416,
165
+ "left_thumb_metacarpal_joint": 1.1781593561172485,
166
+ "left_thumb_proximal_joint": 0.13173270225524902,
167
+ "right_index_distal_joint": 2.8387556194608976e-10,
168
+ "right_index_proximal_joint": 0.0015330095775425434,
169
+ "right_middle_distal_joint": 0.007450763136148453,
170
+ "right_middle_proximal_joint": 0.011103007942438126,
171
+ "right_pinky_distal_joint": 0.0035635153762996197,
172
+ "right_pinky_proximal_joint": 0.5900799632072449,
173
+ "right_ring_distal_joint": 0.005387734156101942,
174
+ "right_ring_proximal_joint": 0.003577441442757845,
175
+ "right_thumb_proximal_joint": 0.15371644496917725
176
+ },
177
+ "coverage": {
178
+ "hard_limit_source": {
179
+ "available": 50,
180
+ "unavailable": 0,
181
+ "unknown": 0
182
+ },
183
+ "joint_role_manifest": {
184
+ "available": 50,
185
+ "unavailable": 0,
186
+ "unknown": 0
187
+ }
188
+ },
189
+ "hard_limit_sources": {
190
+ "runtime.joint_pos_limits": 50
191
+ },
192
+ "soft_limit_sources": {
193
+ "runtime.soft_joint_pos_limits": 50
194
+ },
195
+ "unknown_runtime_joint_names": [],
196
+ "baseline_checks": {
197
+ "post_home": {
198
+ "checked": 50,
199
+ "hard_violation_episodes": 0
200
+ },
201
+ "post_reset": {
202
+ "checked": 50,
203
+ "hard_violation_episodes": 0
204
+ }
205
+ }
206
+ }
207
+ },
208
+ "robot_motion": {
209
+ "joint_acceleration_rms": 15.471890026451332,
210
+ "joint_effort_rms": 25.641329946828304,
211
+ "joint_jerk_rms": 1579.9858726495736,
212
+ "joint_velocity_rms": 0.36679375898850103
213
+ },
214
+ "grasp": {
215
+ "kinematic_grasp_stability_index": 0.8964709626972839,
216
+ "mean_kinematic_grasp_stability": 0.3550903259402908,
217
+ "per_object_best_kinematic_grasp_stability": {
218
+ "obj_025_mug_1": 0.8170701661754,
219
+ "obj_031_spoon_1": 0.7937650808003943
220
+ },
221
+ "grasp_gsi_diagnostics": {
222
+ "mean_anti_slip_score": 0.6460994635604614,
223
+ "mean_duration_score": 0.6828063164901458,
224
+ "mean_hold_score": 0.405139165747582,
225
+ "mean_kinematic_stability_score": 0.11010418817254077,
226
+ "mean_legacy_gsi": 0.5025088936224401,
227
+ "mean_lift_score": 0.5480513786842346,
228
+ "mean_motion_stability_score": 0.37623352744541655
229
+ }
230
+ },
231
+ "tool": {
232
+ "tool_selection_accuracy": null,
233
+ "tool_switch_success_rate": null,
234
+ "tool_switch_total": 0,
235
+ "tool_switch_successful": 0
236
+ },
237
+ "runtime": {
238
+ "avg_steps": 2580.78,
239
+ "error_count": 0
240
+ }
241
+ }
242
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/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/67_faucet_cup_water_fill/Full/native_shards/000/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 faucet_recovery-ee8f8b886b",
5
+ "base_seed": 206700000,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.65
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 41.08452380952276
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.56,
64
+ "hard_violation_rate": 0.04,
65
+ "drop_rate": 0.04,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 1.5077725675859054e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.56,
74
+ "at_end_success_rate": 0.56,
75
+ "task_family_success_rate": {
76
+ "articulation_multi": 0.56
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.5,
81
+ "current_chain_depth_progress_score": 0.65,
82
+ "latched_chain_depth_progress_score": 0.65,
83
+ "avg_current_chain_depth": 2.6,
84
+ "avg_latched_chain_depth": 2.6
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.785287531616784,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2465.0714285714284,
91
+ "avg_policy_steps_to_success": 822.0714285714286,
92
+ "avg_policy_queries_to_success": 822.0714285714286
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 1.5077725675859054e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_031_spoon_1": 1
98
+ },
99
+ "robot_constraint_violation_rate": 0.96,
100
+ "finger_joint_limit_saturation_rate": 0.4,
101
+ "joint_limit_active_step_rate": 0.0971005533525323,
102
+ "finger_joint_limit_saturation_step_rate": 0.04074001477617116,
103
+ "max_joint_limit_excess_rad": 0.6413490772247314,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {
106
+ "l_joint4": 33,
107
+ "left_index_proximal_joint": 648,
108
+ "left_middle_distal_joint": 28,
109
+ "left_middle_proximal_joint": 3204,
110
+ "left_pinky_distal_joint": 75,
111
+ "left_pinky_proximal_joint": 459,
112
+ "left_ring_distal_joint": 98,
113
+ "left_ring_proximal_joint": 4823,
114
+ "left_thumb_distal_joint": 164
115
+ },
116
+ "joint_limit_hard_counts": {
117
+ "l_joint4": 1,
118
+ "left_index_proximal_joint": 213,
119
+ "left_middle_distal_joint": 43,
120
+ "left_middle_proximal_joint": 76,
121
+ "left_pinky_distal_joint": 29,
122
+ "left_pinky_proximal_joint": 296,
123
+ "left_ring_distal_joint": 40,
124
+ "left_ring_proximal_joint": 186,
125
+ "left_thumb_distal_joint": 8,
126
+ "right_pinky_proximal_joint": 14
127
+ },
128
+ "joint_limit_hard_episode_events": 498,
129
+ "finger_joint_limit_saturation_counts": {
130
+ "left_index_proximal_joint": 230,
131
+ "left_middle_proximal_joint": 2319,
132
+ "left_pinky_distal_joint": 3,
133
+ "left_pinky_proximal_joint": 28,
134
+ "left_ring_distal_joint": 25,
135
+ "left_ring_proximal_joint": 2188,
136
+ "left_thumb_distal_joint": 70
137
+ },
138
+ "joint_limit_max_excess_by_joint": {
139
+ "l_joint4": 0.19756436347961426,
140
+ "left_index_distal_joint": 1.4368410461429448e-08,
141
+ "left_index_proximal_joint": 0.38475552201271057,
142
+ "left_middle_distal_joint": 0.10501216351985931,
143
+ "left_middle_proximal_joint": 0.5079748630523682,
144
+ "left_pinky_distal_joint": 0.23223185539245605,
145
+ "left_pinky_proximal_joint": 0.6413490772247314,
146
+ "left_ring_distal_joint": 0.2590444087982178,
147
+ "left_ring_proximal_joint": 0.60569167137146,
148
+ "left_thumb_distal_joint": 0.5440137386322021,
149
+ "left_thumb_metacarpal_joint": 0.000139108145958744,
150
+ "left_thumb_proximal_joint": 0.001184268156066537,
151
+ "right_index_distal_joint": 2.876058002865278e-10,
152
+ "right_index_proximal_joint": 0.0015717557398602366,
153
+ "right_middle_distal_joint": 0.00484284944832325,
154
+ "right_middle_proximal_joint": 0.004668414127081633,
155
+ "right_pinky_distal_joint": 0.0020061666145920753,
156
+ "right_pinky_proximal_joint": 0.005820153746753931,
157
+ "right_ring_distal_joint": 0.003391790669411421,
158
+ "right_ring_proximal_joint": 0.0034554610028862953,
159
+ "right_thumb_proximal_joint": 0.0003381705901119858
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": 14.246761581769015,
194
+ "joint_effort_rms": 27.43918545809793,
195
+ "joint_jerk_rms": 1456.9715999975804,
196
+ "joint_velocity_rms": 0.32792123575860815
197
+ },
198
+ "grasp": {
199
+ "kinematic_grasp_stability_index": 0.7721357059626545,
200
+ "mean_kinematic_grasp_stability": 0.34967989919677034,
201
+ "per_object_best_kinematic_grasp_stability": {
202
+ "obj_025_mug_1": 0.6083930895157569,
203
+ "obj_031_spoon_1": 0.6990438208954461
204
+ },
205
+ "grasp_gsi_diagnostics": {
206
+ "mean_anti_slip_score": 0.6984368167843593,
207
+ "mean_duration_score": 0.5413306820864666,
208
+ "mean_hold_score": 0.3441717789686351,
209
+ "mean_kinematic_stability_score": 0.07836672012058572,
210
+ "mean_legacy_gsi": 0.44640693831369394,
211
+ "mean_lift_score": 0.4532624441951466,
212
+ "mean_motion_stability_score": 0.31512160624969865
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": 2652.92,
223
+ "error_count": 0
224
+ }
225
+ }
226
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/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/67_faucet_cup_water_fill/Full/native_shards/001/summary.json ADDED
@@ -0,0 +1,233 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA faucet_recovery-ee8f8b886b",
5
+ "base_seed": 206700000,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.7
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 40.237499999999
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.48,
64
+ "hard_violation_rate": 0.12,
65
+ "drop_rate": 0.12,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 6.008983430228191e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.48,
74
+ "at_end_success_rate": 0.48,
75
+ "task_family_success_rate": {
76
+ "articulation_multi": 0.48
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.51,
81
+ "current_chain_depth_progress_score": 0.67,
82
+ "latched_chain_depth_progress_score": 0.7,
83
+ "avg_current_chain_depth": 2.68,
84
+ "avg_latched_chain_depth": 2.8
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.800931379375414,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2414.25,
91
+ "avg_policy_steps_to_success": 805.0833333333334,
92
+ "avg_policy_queries_to_success": 805.0833333333334
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 6.008983430228191e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_025_mug_1": 1,
98
+ "drop_to_table_obj_031_spoon_1": 3
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.56,
102
+ "joint_limit_active_step_rate": 0.022173148857542024,
103
+ "finger_joint_limit_saturation_step_rate": 0.007586341580663091,
104
+ "max_joint_limit_excess_rad": 0.7685144543647766,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "left_index_proximal_joint": 990,
108
+ "left_middle_distal_joint": 106,
109
+ "left_middle_proximal_joint": 48,
110
+ "left_pinky_distal_joint": 64,
111
+ "left_pinky_proximal_joint": 94,
112
+ "left_ring_distal_joint": 98,
113
+ "left_ring_proximal_joint": 147,
114
+ "left_thumb_distal_joint": 79,
115
+ "left_thumb_metacarpal_joint": 113,
116
+ "left_thumb_proximal_joint": 3,
117
+ "right_pinky_proximal_joint": 30
118
+ },
119
+ "joint_limit_hard_counts": {
120
+ "left_index_proximal_joint": 308,
121
+ "left_middle_distal_joint": 60,
122
+ "left_middle_proximal_joint": 48,
123
+ "left_pinky_distal_joint": 68,
124
+ "left_pinky_proximal_joint": 220,
125
+ "left_ring_distal_joint": 99,
126
+ "left_ring_proximal_joint": 100,
127
+ "left_thumb_distal_joint": 7,
128
+ "left_thumb_metacarpal_joint": 5,
129
+ "left_thumb_proximal_joint": 1,
130
+ "right_middle_distal_joint": 5,
131
+ "right_middle_proximal_joint": 6,
132
+ "right_pinky_proximal_joint": 26,
133
+ "right_ring_distal_joint": 1,
134
+ "right_ring_proximal_joint": 1
135
+ },
136
+ "joint_limit_hard_episode_events": 660,
137
+ "finger_joint_limit_saturation_counts": {
138
+ "left_index_proximal_joint": 396,
139
+ "left_middle_distal_joint": 20,
140
+ "left_middle_proximal_joint": 4,
141
+ "left_pinky_proximal_joint": 28,
142
+ "left_ring_proximal_joint": 32,
143
+ "left_thumb_distal_joint": 23,
144
+ "left_thumb_metacarpal_joint": 81
145
+ },
146
+ "joint_limit_max_excess_by_joint": {
147
+ "left_index_distal_joint": 2.1604385835871653e-08,
148
+ "left_index_proximal_joint": 0.7220959663391113,
149
+ "left_middle_distal_joint": 0.5097897052764893,
150
+ "left_middle_proximal_joint": 0.301624059677124,
151
+ "left_pinky_distal_joint": 0.2474699318408966,
152
+ "left_pinky_proximal_joint": 0.646672248840332,
153
+ "left_ring_distal_joint": 0.17256319522857666,
154
+ "left_ring_proximal_joint": 0.7269802093505859,
155
+ "left_thumb_distal_joint": 0.6144053936004639,
156
+ "left_thumb_metacarpal_joint": 0.7685144543647766,
157
+ "left_thumb_proximal_joint": 0.07786798477172852,
158
+ "right_index_distal_joint": 2.826884282214337e-10,
159
+ "right_index_proximal_joint": 0.0016674681100994349,
160
+ "right_middle_distal_joint": 0.008010623045265675,
161
+ "right_middle_proximal_joint": 0.005687232129275799,
162
+ "right_pinky_distal_joint": 0.003878065850585699,
163
+ "right_pinky_proximal_joint": 0.06665533035993576,
164
+ "right_ring_distal_joint": 0.0062949275597929955,
165
+ "right_ring_proximal_joint": 0.006654907017946243,
166
+ "right_thumb_proximal_joint": 0.00033610541140660644
167
+ },
168
+ "coverage": {
169
+ "hard_limit_source": {
170
+ "available": 25,
171
+ "unavailable": 0,
172
+ "unknown": 0
173
+ },
174
+ "joint_role_manifest": {
175
+ "available": 25,
176
+ "unavailable": 0,
177
+ "unknown": 0
178
+ }
179
+ },
180
+ "hard_limit_sources": {
181
+ "runtime.joint_pos_limits": 25
182
+ },
183
+ "soft_limit_sources": {
184
+ "runtime.soft_joint_pos_limits": 25
185
+ },
186
+ "unknown_runtime_joint_names": [],
187
+ "baseline_checks": {
188
+ "post_home": {
189
+ "checked": 25,
190
+ "hard_violation_episodes": 0
191
+ },
192
+ "post_reset": {
193
+ "checked": 25,
194
+ "hard_violation_episodes": 0
195
+ }
196
+ }
197
+ }
198
+ },
199
+ "robot_motion": {
200
+ "joint_acceleration_rms": 15.761108624213113,
201
+ "joint_effort_rms": 24.376740323626894,
202
+ "joint_jerk_rms": 1616.3904896939816,
203
+ "joint_velocity_rms": 0.36840860878515935
204
+ },
205
+ "grasp": {
206
+ "kinematic_grasp_stability_index": 0.8860186306014868,
207
+ "mean_kinematic_grasp_stability": 0.3750686595861236,
208
+ "per_object_best_kinematic_grasp_stability": {
209
+ "obj_025_mug_1": 0.7759971660695009,
210
+ "obj_031_spoon_1": 0.7584216034551822
211
+ },
212
+ "grasp_gsi_diagnostics": {
213
+ "mean_anti_slip_score": 0.6517598600713672,
214
+ "mean_duration_score": 0.6815891832166516,
215
+ "mean_hold_score": 0.395778563424798,
216
+ "mean_kinematic_stability_score": 0.1145681237984081,
217
+ "mean_legacy_gsi": 0.4977794305659731,
218
+ "mean_lift_score": 0.5299775244875361,
219
+ "mean_motion_stability_score": 0.36794796568646504
220
+ }
221
+ },
222
+ "tool": {
223
+ "tool_selection_accuracy": null,
224
+ "tool_switch_success_rate": null,
225
+ "tool_switch_total": 0,
226
+ "tool_switch_successful": 0
227
+ },
228
+ "runtime": {
229
+ "avg_steps": 2662.68,
230
+ "error_count": 0
231
+ }
232
+ }
233
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Full/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Full/summary.json ADDED
@@ -0,0 +1,238 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA faucet_recovery-ee8f8b886b",
5
+ "base_seed": 206700000,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.675
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 40.69358974358872
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.52,
64
+ "hard_violation_rate": 0.08,
65
+ "drop_rate": 0.08,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 3.762510346903454e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.52,
74
+ "at_end_success_rate": 0.52,
75
+ "task_family_success_rate": {
76
+ "articulation_multi": 0.52
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.505,
81
+ "current_chain_depth_progress_score": 0.66,
82
+ "latched_chain_depth_progress_score": 0.675,
83
+ "avg_current_chain_depth": 2.64,
84
+ "avg_latched_chain_depth": 2.7
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.792507769043844,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2441.6153846153848,
91
+ "avg_policy_steps_to_success": 814.2307692307693,
92
+ "avg_policy_queries_to_success": 814.2307692307693
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 3.762510346903454e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_025_mug_1": 1,
98
+ "drop_to_table_obj_031_spoon_1": 4
99
+ },
100
+ "robot_constraint_violation_rate": 0.98,
101
+ "finger_joint_limit_saturation_rate": 0.48,
102
+ "joint_limit_active_step_rate": 0.059568063812175484,
103
+ "finger_joint_limit_saturation_step_rate": 0.024132741365038755,
104
+ "max_joint_limit_excess_rad": 0.7685144543647766,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "l_joint4": 33,
108
+ "left_index_proximal_joint": 1638,
109
+ "left_middle_distal_joint": 134,
110
+ "left_middle_proximal_joint": 3252,
111
+ "left_pinky_distal_joint": 139,
112
+ "left_pinky_proximal_joint": 553,
113
+ "left_ring_distal_joint": 196,
114
+ "left_ring_proximal_joint": 4970,
115
+ "left_thumb_distal_joint": 243,
116
+ "left_thumb_metacarpal_joint": 113,
117
+ "left_thumb_proximal_joint": 3,
118
+ "right_pinky_proximal_joint": 30
119
+ },
120
+ "joint_limit_hard_counts": {
121
+ "l_joint4": 1,
122
+ "left_index_proximal_joint": 521,
123
+ "left_middle_distal_joint": 103,
124
+ "left_middle_proximal_joint": 124,
125
+ "left_pinky_distal_joint": 97,
126
+ "left_pinky_proximal_joint": 516,
127
+ "left_ring_distal_joint": 139,
128
+ "left_ring_proximal_joint": 286,
129
+ "left_thumb_distal_joint": 15,
130
+ "left_thumb_metacarpal_joint": 5,
131
+ "left_thumb_proximal_joint": 1,
132
+ "right_middle_distal_joint": 5,
133
+ "right_middle_proximal_joint": 6,
134
+ "right_pinky_proximal_joint": 40,
135
+ "right_ring_distal_joint": 1,
136
+ "right_ring_proximal_joint": 1
137
+ },
138
+ "joint_limit_hard_episode_events": 1158,
139
+ "finger_joint_limit_saturation_counts": {
140
+ "left_index_proximal_joint": 626,
141
+ "left_middle_distal_joint": 20,
142
+ "left_middle_proximal_joint": 2323,
143
+ "left_pinky_distal_joint": 3,
144
+ "left_pinky_proximal_joint": 56,
145
+ "left_ring_distal_joint": 25,
146
+ "left_ring_proximal_joint": 2220,
147
+ "left_thumb_distal_joint": 93,
148
+ "left_thumb_metacarpal_joint": 81
149
+ },
150
+ "joint_limit_max_excess_by_joint": {
151
+ "l_joint4": 0.19756436347961426,
152
+ "left_index_distal_joint": 2.1604385835871653e-08,
153
+ "left_index_proximal_joint": 0.7220959663391113,
154
+ "left_middle_distal_joint": 0.5097897052764893,
155
+ "left_middle_proximal_joint": 0.5079748630523682,
156
+ "left_pinky_distal_joint": 0.2474699318408966,
157
+ "left_pinky_proximal_joint": 0.646672248840332,
158
+ "left_ring_distal_joint": 0.2590444087982178,
159
+ "left_ring_proximal_joint": 0.7269802093505859,
160
+ "left_thumb_distal_joint": 0.6144053936004639,
161
+ "left_thumb_metacarpal_joint": 0.7685144543647766,
162
+ "left_thumb_proximal_joint": 0.07786798477172852,
163
+ "right_index_distal_joint": 2.876058002865278e-10,
164
+ "right_index_proximal_joint": 0.0016674681100994349,
165
+ "right_middle_distal_joint": 0.008010623045265675,
166
+ "right_middle_proximal_joint": 0.005687232129275799,
167
+ "right_pinky_distal_joint": 0.003878065850585699,
168
+ "right_pinky_proximal_joint": 0.06665533035993576,
169
+ "right_ring_distal_joint": 0.0062949275597929955,
170
+ "right_ring_proximal_joint": 0.006654907017946243,
171
+ "right_thumb_proximal_joint": 0.0003381705901119858
172
+ },
173
+ "coverage": {
174
+ "hard_limit_source": {
175
+ "available": 50,
176
+ "unavailable": 0,
177
+ "unknown": 0
178
+ },
179
+ "joint_role_manifest": {
180
+ "available": 50,
181
+ "unavailable": 0,
182
+ "unknown": 0
183
+ }
184
+ },
185
+ "hard_limit_sources": {
186
+ "runtime.joint_pos_limits": 50
187
+ },
188
+ "soft_limit_sources": {
189
+ "runtime.soft_joint_pos_limits": 50
190
+ },
191
+ "unknown_runtime_joint_names": [],
192
+ "baseline_checks": {
193
+ "post_home": {
194
+ "checked": 50,
195
+ "hard_violation_episodes": 0
196
+ },
197
+ "post_reset": {
198
+ "checked": 50,
199
+ "hard_violation_episodes": 0
200
+ }
201
+ }
202
+ }
203
+ },
204
+ "robot_motion": {
205
+ "joint_acceleration_rms": 15.003935102991065,
206
+ "joint_effort_rms": 25.90796289086241,
207
+ "joint_jerk_rms": 1536.681044845781,
208
+ "joint_velocity_rms": 0.3481649222718838
209
+ },
210
+ "grasp": {
211
+ "kinematic_grasp_stability_index": 0.8290771682820706,
212
+ "mean_kinematic_grasp_stability": 0.36318455897834123,
213
+ "per_object_best_kinematic_grasp_stability": {
214
+ "obj_025_mug_1": 0.6921951277926289,
215
+ "obj_031_spoon_1": 0.7287327121753142
216
+ },
217
+ "grasp_gsi_diagnostics": {
218
+ "mean_anti_slip_score": 0.6750983384278633,
219
+ "mean_duration_score": 0.6114599326515592,
220
+ "mean_hold_score": 0.3699751711967166,
221
+ "mean_kinematic_stability_score": 0.09646742195949691,
222
+ "mean_legacy_gsi": 0.47209318443983356,
223
+ "mean_lift_score": 0.49161998434134135,
224
+ "mean_motion_stability_score": 0.34153478596808184
225
+ }
226
+ },
227
+ "tool": {
228
+ "tool_selection_accuracy": null,
229
+ "tool_switch_success_rate": null,
230
+ "tool_switch_total": 0,
231
+ "tool_switch_successful": 0
232
+ },
233
+ "runtime": {
234
+ "avg_steps": 2657.8,
235
+ "error_count": 0
236
+ }
237
+ }
238
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/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/67_faucet_cup_water_fill/Inv/native_shards/000/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 faucet_recovery-ee8f8b886b",
5
+ "base_seed": 106700000,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.74
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 40.511458333332314
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": 7.701905451408678e-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
+ "articulation_multi": 0.64
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.53,
81
+ "current_chain_depth_progress_score": 0.71,
82
+ "latched_chain_depth_progress_score": 0.74,
83
+ "avg_current_chain_depth": 2.84,
84
+ "avg_latched_chain_depth": 2.96
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7972713297672812,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2430.6875,
91
+ "avg_policy_steps_to_success": 810.5625,
92
+ "avg_policy_queries_to_success": 810.5625
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 7.701905451408678e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_031_spoon_1": 5
98
+ },
99
+ "robot_constraint_violation_rate": 1.0,
100
+ "finger_joint_limit_saturation_rate": 0.4,
101
+ "joint_limit_active_step_rate": 0.018561592137894916,
102
+ "finger_joint_limit_saturation_step_rate": 0.00839507694203546,
103
+ "max_joint_limit_excess_rad": 1.7332388162612915,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {
106
+ "l_joint4": 41,
107
+ "left_index_proximal_joint": 735,
108
+ "left_middle_distal_joint": 81,
109
+ "left_middle_proximal_joint": 36,
110
+ "left_pinky_distal_joint": 45,
111
+ "left_pinky_proximal_joint": 49,
112
+ "left_ring_distal_joint": 70,
113
+ "left_ring_proximal_joint": 33,
114
+ "left_thumb_distal_joint": 123,
115
+ "left_thumb_metacarpal_joint": 176,
116
+ "left_thumb_proximal_joint": 6,
117
+ "right_pinky_proximal_joint": 11
118
+ },
119
+ "joint_limit_hard_counts": {
120
+ "l_joint4": 1,
121
+ "left_index_proximal_joint": 158,
122
+ "left_middle_distal_joint": 58,
123
+ "left_middle_proximal_joint": 45,
124
+ "left_pinky_distal_joint": 49,
125
+ "left_pinky_proximal_joint": 185,
126
+ "left_ring_distal_joint": 65,
127
+ "left_ring_proximal_joint": 84,
128
+ "left_thumb_distal_joint": 7,
129
+ "left_thumb_metacarpal_joint": 3,
130
+ "left_thumb_proximal_joint": 1,
131
+ "right_middle_proximal_joint": 3,
132
+ "right_pinky_proximal_joint": 25,
133
+ "right_ring_proximal_joint": 2
134
+ },
135
+ "joint_limit_hard_episode_events": 433,
136
+ "finger_joint_limit_saturation_counts": {
137
+ "left_index_proximal_joint": 327,
138
+ "left_middle_distal_joint": 27,
139
+ "left_middle_proximal_joint": 3,
140
+ "left_thumb_distal_joint": 50,
141
+ "left_thumb_metacarpal_joint": 164
142
+ },
143
+ "joint_limit_max_excess_by_joint": {
144
+ "l_joint4": 0.26824164390563965,
145
+ "left_index_distal_joint": 9.520997323875235e-09,
146
+ "left_index_proximal_joint": 0.6279017329216003,
147
+ "left_middle_distal_joint": 1.060793399810791,
148
+ "left_middle_proximal_joint": 0.2891428470611572,
149
+ "left_pinky_distal_joint": 0.07634079456329346,
150
+ "left_pinky_proximal_joint": 0.053174734115600586,
151
+ "left_ring_distal_joint": 0.08001899719238281,
152
+ "left_ring_proximal_joint": 0.05351388454437256,
153
+ "left_thumb_distal_joint": 1.0086126327514648,
154
+ "left_thumb_metacarpal_joint": 1.7332388162612915,
155
+ "left_thumb_proximal_joint": 0.10967326164245605,
156
+ "right_index_distal_joint": 2.858066006083959e-10,
157
+ "right_index_proximal_joint": 0.0016676164232194424,
158
+ "right_middle_distal_joint": 0.0039688590914011,
159
+ "right_middle_proximal_joint": 0.005757241044193506,
160
+ "right_pinky_distal_joint": 0.0017661906313151121,
161
+ "right_pinky_proximal_joint": 0.08219242095947266,
162
+ "right_ring_distal_joint": 0.003288484178483486,
163
+ "right_ring_proximal_joint": 0.005109657999128103,
164
+ "right_thumb_proximal_joint": 0.0003391732170712203
165
+ },
166
+ "coverage": {
167
+ "hard_limit_source": {
168
+ "available": 25,
169
+ "unavailable": 0,
170
+ "unknown": 0
171
+ },
172
+ "joint_role_manifest": {
173
+ "available": 25,
174
+ "unavailable": 0,
175
+ "unknown": 0
176
+ }
177
+ },
178
+ "hard_limit_sources": {
179
+ "runtime.joint_pos_limits": 25
180
+ },
181
+ "soft_limit_sources": {
182
+ "runtime.soft_joint_pos_limits": 25
183
+ },
184
+ "unknown_runtime_joint_names": [],
185
+ "baseline_checks": {
186
+ "post_home": {
187
+ "checked": 25,
188
+ "hard_violation_episodes": 0
189
+ },
190
+ "post_reset": {
191
+ "checked": 25,
192
+ "hard_violation_episodes": 0
193
+ }
194
+ }
195
+ }
196
+ },
197
+ "robot_motion": {
198
+ "joint_acceleration_rms": 14.63532806836195,
199
+ "joint_effort_rms": 26.360950503048976,
200
+ "joint_jerk_rms": 1499.0225365498031,
201
+ "joint_velocity_rms": 0.3494872756109929
202
+ },
203
+ "grasp": {
204
+ "kinematic_grasp_stability_index": 0.9105282435897784,
205
+ "mean_kinematic_grasp_stability": 0.32119002383585105,
206
+ "per_object_best_kinematic_grasp_stability": {
207
+ "obj_025_mug_1": 0.76389673395113,
208
+ "obj_031_spoon_1": 0.7867560059886339
209
+ },
210
+ "grasp_gsi_diagnostics": {
211
+ "mean_anti_slip_score": 0.6685838429959361,
212
+ "mean_duration_score": 0.6173789532222063,
213
+ "mean_hold_score": 0.34686956714328965,
214
+ "mean_kinematic_stability_score": 0.07870473981469867,
215
+ "mean_legacy_gsi": 0.4611471480997945,
216
+ "mean_lift_score": 0.47724596073956776,
217
+ "mean_motion_stability_score": 0.3228933673617849
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": 2596.76,
228
+ "error_count": 0
229
+ }
230
+ }
231
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/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/67_faucet_cup_water_fill/Inv/native_shards/001/summary.json ADDED
@@ -0,0 +1,236 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA faucet_recovery-ee8f8b886b",
5
+ "base_seed": 106700000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 15,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.6,
50
+ "success_rate_ci": {
51
+ "mean": 0.6,
52
+ "ci_low": 0.4,
53
+ "ci_high": 0.8
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.74
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 42.34333333333221
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.6,
64
+ "hard_violation_rate": 0.12,
65
+ "drop_rate": 0.12,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 4.4756747079622256e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.6,
74
+ "at_end_success_rate": 0.6,
75
+ "task_family_success_rate": {
76
+ "articulation_multi": 0.6
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.54,
81
+ "current_chain_depth_progress_score": 0.74,
82
+ "latched_chain_depth_progress_score": 0.74,
83
+ "avg_current_chain_depth": 2.96,
84
+ "avg_latched_chain_depth": 2.96
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7611703248631769,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2540.6,
91
+ "avg_policy_steps_to_success": 847.3333333333334,
92
+ "avg_policy_queries_to_success": 847.3333333333334
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 4.4756747079622256e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_031_spoon_1": 3
98
+ },
99
+ "robot_constraint_violation_rate": 0.96,
100
+ "finger_joint_limit_saturation_rate": 0.48,
101
+ "joint_limit_active_step_rate": 0.02021513076429605,
102
+ "finger_joint_limit_saturation_step_rate": 0.007608647003535783,
103
+ "max_joint_limit_excess_rad": 1.4746557474136353,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {
106
+ "l_joint4": 54,
107
+ "left_index_proximal_joint": 600,
108
+ "left_middle_distal_joint": 90,
109
+ "left_middle_proximal_joint": 41,
110
+ "left_pinky_distal_joint": 136,
111
+ "left_pinky_proximal_joint": 73,
112
+ "left_ring_distal_joint": 155,
113
+ "left_ring_proximal_joint": 93,
114
+ "left_thumb_distal_joint": 93,
115
+ "left_thumb_metacarpal_joint": 272,
116
+ "left_thumb_proximal_joint": 18,
117
+ "right_middle_proximal_joint": 1,
118
+ "right_pinky_proximal_joint": 151
119
+ },
120
+ "joint_limit_hard_counts": {
121
+ "l_joint4": 1,
122
+ "left_index_proximal_joint": 431,
123
+ "left_middle_distal_joint": 58,
124
+ "left_middle_proximal_joint": 27,
125
+ "left_pinky_distal_joint": 48,
126
+ "left_pinky_proximal_joint": 177,
127
+ "left_ring_distal_joint": 85,
128
+ "left_ring_proximal_joint": 74,
129
+ "left_thumb_distal_joint": 6,
130
+ "left_thumb_metacarpal_joint": 5,
131
+ "left_thumb_proximal_joint": 3,
132
+ "right_middle_distal_joint": 1,
133
+ "right_middle_proximal_joint": 8,
134
+ "right_pinky_proximal_joint": 21,
135
+ "right_ring_proximal_joint": 5
136
+ },
137
+ "joint_limit_hard_episode_events": 731,
138
+ "finger_joint_limit_saturation_counts": {
139
+ "left_index_proximal_joint": 99,
140
+ "left_middle_proximal_joint": 18,
141
+ "left_pinky_distal_joint": 63,
142
+ "left_ring_distal_joint": 56,
143
+ "left_ring_proximal_joint": 54,
144
+ "left_thumb_distal_joint": 35,
145
+ "left_thumb_metacarpal_joint": 250,
146
+ "right_pinky_proximal_joint": 98
147
+ },
148
+ "joint_limit_max_excess_by_joint": {
149
+ "l_joint4": 0.061086416244506836,
150
+ "left_index_distal_joint": 1.21904237815329e-08,
151
+ "left_index_proximal_joint": 0.4759722054004669,
152
+ "left_middle_distal_joint": 0.2082914113998413,
153
+ "left_middle_proximal_joint": 0.6001198291778564,
154
+ "left_pinky_distal_joint": 0.31313037872314453,
155
+ "left_pinky_proximal_joint": 0.18728435039520264,
156
+ "left_ring_distal_joint": 0.26525425910949707,
157
+ "left_ring_proximal_joint": 0.7280724048614502,
158
+ "left_thumb_distal_joint": 0.7790330648422241,
159
+ "left_thumb_metacarpal_joint": 1.4746557474136353,
160
+ "left_thumb_proximal_joint": 0.19806838035583496,
161
+ "right_index_distal_joint": 2.876058002865278e-10,
162
+ "right_index_proximal_joint": 0.001565468031913042,
163
+ "right_middle_distal_joint": 0.005032039247453213,
164
+ "right_middle_proximal_joint": 0.010981420986354351,
165
+ "right_pinky_distal_joint": 0.0023103938437998295,
166
+ "right_pinky_proximal_joint": 0.4234847128391266,
167
+ "right_ring_distal_joint": 0.0037545976229012012,
168
+ "right_ring_proximal_joint": 0.008314892649650574,
169
+ "right_thumb_proximal_joint": 0.00034283031709492207
170
+ },
171
+ "coverage": {
172
+ "hard_limit_source": {
173
+ "available": 25,
174
+ "unavailable": 0,
175
+ "unknown": 0
176
+ },
177
+ "joint_role_manifest": {
178
+ "available": 25,
179
+ "unavailable": 0,
180
+ "unknown": 0
181
+ }
182
+ },
183
+ "hard_limit_sources": {
184
+ "runtime.joint_pos_limits": 25
185
+ },
186
+ "soft_limit_sources": {
187
+ "runtime.soft_joint_pos_limits": 25
188
+ },
189
+ "unknown_runtime_joint_names": [],
190
+ "baseline_checks": {
191
+ "post_home": {
192
+ "checked": 25,
193
+ "hard_violation_episodes": 0
194
+ },
195
+ "post_reset": {
196
+ "checked": 25,
197
+ "hard_violation_episodes": 0
198
+ }
199
+ }
200
+ }
201
+ },
202
+ "robot_motion": {
203
+ "joint_acceleration_rms": 14.696777707783028,
204
+ "joint_effort_rms": 24.83982024391362,
205
+ "joint_jerk_rms": 1498.5023211345308,
206
+ "joint_velocity_rms": 0.35444284657295505
207
+ },
208
+ "grasp": {
209
+ "kinematic_grasp_stability_index": 0.819651481614462,
210
+ "mean_kinematic_grasp_stability": 0.3625249568594325,
211
+ "per_object_best_kinematic_grasp_stability": {
212
+ "obj_025_mug_1": 0.7237064584394186,
213
+ "obj_031_spoon_1": 0.734126028144652
214
+ },
215
+ "grasp_gsi_diagnostics": {
216
+ "mean_anti_slip_score": 0.6642841705199767,
217
+ "mean_duration_score": 0.6300499406922953,
218
+ "mean_hold_score": 0.3526808786920836,
219
+ "mean_kinematic_stability_score": 0.07998442908138317,
220
+ "mean_legacy_gsi": 0.46391799179206944,
221
+ "mean_lift_score": 0.467448242028829,
222
+ "mean_motion_stability_score": 0.32608787490143337
223
+ }
224
+ },
225
+ "tool": {
226
+ "tool_selection_accuracy": null,
227
+ "tool_switch_success_rate": null,
228
+ "tool_switch_total": 0,
229
+ "tool_switch_successful": 0
230
+ },
231
+ "runtime": {
232
+ "avg_steps": 2681.16,
233
+ "error_count": 0
234
+ }
235
+ }
236
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Inv/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/Inv/summary.json ADDED
@@ -0,0 +1,237 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA faucet_recovery-ee8f8b886b",
5
+ "base_seed": 106700000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 31,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.62,
50
+ "success_rate_ci": {
51
+ "mean": 0.62,
52
+ "ci_low": 0.48,
53
+ "ci_high": 0.74
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.74
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 41.397849462364526
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.62,
64
+ "hard_violation_rate": 0.16,
65
+ "drop_rate": 0.16,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 6.0629945129899655e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.62,
74
+ "at_end_success_rate": 0.62,
75
+ "task_family_success_rate": {
76
+ "articulation_multi": 0.62
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.535,
81
+ "current_chain_depth_progress_score": 0.725,
82
+ "latched_chain_depth_progress_score": 0.74,
83
+ "avg_current_chain_depth": 2.9,
84
+ "avg_latched_chain_depth": 2.96
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7798031015878759,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2483.8709677419356,
91
+ "avg_policy_steps_to_success": 828.3548387096774,
92
+ "avg_policy_queries_to_success": 828.3548387096774
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 6.0629945129899655e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_031_spoon_1": 8
98
+ },
99
+ "robot_constraint_violation_rate": 0.98,
100
+ "finger_joint_limit_saturation_rate": 0.44,
101
+ "joint_limit_active_step_rate": 0.01940158244156789,
102
+ "finger_joint_limit_saturation_step_rate": 0.007995574014005517,
103
+ "max_joint_limit_excess_rad": 1.7332388162612915,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {
106
+ "l_joint4": 95,
107
+ "left_index_proximal_joint": 1335,
108
+ "left_middle_distal_joint": 171,
109
+ "left_middle_proximal_joint": 77,
110
+ "left_pinky_distal_joint": 181,
111
+ "left_pinky_proximal_joint": 122,
112
+ "left_ring_distal_joint": 225,
113
+ "left_ring_proximal_joint": 126,
114
+ "left_thumb_distal_joint": 216,
115
+ "left_thumb_metacarpal_joint": 448,
116
+ "left_thumb_proximal_joint": 24,
117
+ "right_middle_proximal_joint": 1,
118
+ "right_pinky_proximal_joint": 162
119
+ },
120
+ "joint_limit_hard_counts": {
121
+ "l_joint4": 2,
122
+ "left_index_proximal_joint": 589,
123
+ "left_middle_distal_joint": 116,
124
+ "left_middle_proximal_joint": 72,
125
+ "left_pinky_distal_joint": 97,
126
+ "left_pinky_proximal_joint": 362,
127
+ "left_ring_distal_joint": 150,
128
+ "left_ring_proximal_joint": 158,
129
+ "left_thumb_distal_joint": 13,
130
+ "left_thumb_metacarpal_joint": 8,
131
+ "left_thumb_proximal_joint": 4,
132
+ "right_middle_distal_joint": 1,
133
+ "right_middle_proximal_joint": 11,
134
+ "right_pinky_proximal_joint": 46,
135
+ "right_ring_proximal_joint": 7
136
+ },
137
+ "joint_limit_hard_episode_events": 1164,
138
+ "finger_joint_limit_saturation_counts": {
139
+ "left_index_proximal_joint": 426,
140
+ "left_middle_distal_joint": 27,
141
+ "left_middle_proximal_joint": 21,
142
+ "left_pinky_distal_joint": 63,
143
+ "left_ring_distal_joint": 56,
144
+ "left_ring_proximal_joint": 54,
145
+ "left_thumb_distal_joint": 85,
146
+ "left_thumb_metacarpal_joint": 414,
147
+ "right_pinky_proximal_joint": 98
148
+ },
149
+ "joint_limit_max_excess_by_joint": {
150
+ "l_joint4": 0.26824164390563965,
151
+ "left_index_distal_joint": 1.21904237815329e-08,
152
+ "left_index_proximal_joint": 0.6279017329216003,
153
+ "left_middle_distal_joint": 1.060793399810791,
154
+ "left_middle_proximal_joint": 0.6001198291778564,
155
+ "left_pinky_distal_joint": 0.31313037872314453,
156
+ "left_pinky_proximal_joint": 0.18728435039520264,
157
+ "left_ring_distal_joint": 0.26525425910949707,
158
+ "left_ring_proximal_joint": 0.7280724048614502,
159
+ "left_thumb_distal_joint": 1.0086126327514648,
160
+ "left_thumb_metacarpal_joint": 1.7332388162612915,
161
+ "left_thumb_proximal_joint": 0.19806838035583496,
162
+ "right_index_distal_joint": 2.876058002865278e-10,
163
+ "right_index_proximal_joint": 0.0016676164232194424,
164
+ "right_middle_distal_joint": 0.005032039247453213,
165
+ "right_middle_proximal_joint": 0.010981420986354351,
166
+ "right_pinky_distal_joint": 0.0023103938437998295,
167
+ "right_pinky_proximal_joint": 0.4234847128391266,
168
+ "right_ring_distal_joint": 0.0037545976229012012,
169
+ "right_ring_proximal_joint": 0.008314892649650574,
170
+ "right_thumb_proximal_joint": 0.00034283031709492207
171
+ },
172
+ "coverage": {
173
+ "hard_limit_source": {
174
+ "available": 50,
175
+ "unavailable": 0,
176
+ "unknown": 0
177
+ },
178
+ "joint_role_manifest": {
179
+ "available": 50,
180
+ "unavailable": 0,
181
+ "unknown": 0
182
+ }
183
+ },
184
+ "hard_limit_sources": {
185
+ "runtime.joint_pos_limits": 50
186
+ },
187
+ "soft_limit_sources": {
188
+ "runtime.soft_joint_pos_limits": 50
189
+ },
190
+ "unknown_runtime_joint_names": [],
191
+ "baseline_checks": {
192
+ "post_home": {
193
+ "checked": 50,
194
+ "hard_violation_episodes": 0
195
+ },
196
+ "post_reset": {
197
+ "checked": 50,
198
+ "hard_violation_episodes": 0
199
+ }
200
+ }
201
+ }
202
+ },
203
+ "robot_motion": {
204
+ "joint_acceleration_rms": 14.666052888072489,
205
+ "joint_effort_rms": 25.6003853734813,
206
+ "joint_jerk_rms": 1498.762428842167,
207
+ "joint_velocity_rms": 0.351965061091974
208
+ },
209
+ "grasp": {
210
+ "kinematic_grasp_stability_index": 0.8650898626021202,
211
+ "mean_kinematic_grasp_stability": 0.3409963459096505,
212
+ "per_object_best_kinematic_grasp_stability": {
213
+ "obj_025_mug_1": 0.7438015961952743,
214
+ "obj_031_spoon_1": 0.7604410170666429
215
+ },
216
+ "grasp_gsi_diagnostics": {
217
+ "mean_anti_slip_score": 0.6664340067579564,
218
+ "mean_duration_score": 0.6237144469572508,
219
+ "mean_hold_score": 0.3497752229176866,
220
+ "mean_kinematic_stability_score": 0.07934458444804092,
221
+ "mean_legacy_gsi": 0.462532569945932,
222
+ "mean_lift_score": 0.4723471013841984,
223
+ "mean_motion_stability_score": 0.32449062113160915
224
+ }
225
+ },
226
+ "tool": {
227
+ "tool_selection_accuracy": null,
228
+ "tool_switch_success_rate": null,
229
+ "tool_switch_total": 0,
230
+ "tool_switch_successful": 0
231
+ },
232
+ "runtime": {
233
+ "avg_steps": 2638.96,
234
+ "error_count": 0
235
+ }
236
+ }
237
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/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/67_faucet_cup_water_fill/None/native_shards/000/summary.json ADDED
@@ -0,0 +1,236 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA faucet_recovery-ee8f8b886b",
5
+ "base_seed": 106700000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 26,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.5,
50
+ "success_rate_ci": {
51
+ "mean": 0.5,
52
+ "ci_low": 0.3076923076923077,
53
+ "ci_high": 0.6923076923076923
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.6923076923076923
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 41.12948717948613
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.5,
64
+ "hard_violation_rate": 0.07692307692307693,
65
+ "drop_rate": 0.07692307692307693,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 2.870387645851572e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.5,
74
+ "at_end_success_rate": 0.5,
75
+ "task_family_success_rate": {
76
+ "articulation_multi": 0.5
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.5192307692307693,
81
+ "current_chain_depth_progress_score": 0.6730769230769231,
82
+ "latched_chain_depth_progress_score": 0.6923076923076923,
83
+ "avg_current_chain_depth": 2.6923076923076925,
84
+ "avg_latched_chain_depth": 2.769230769230769
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7858596769292344,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2467.769230769231,
91
+ "avg_policy_steps_to_success": 822.9230769230769,
92
+ "avg_policy_queries_to_success": 822.9230769230769
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 2.870387645851572e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_031_spoon_1": 2
98
+ },
99
+ "robot_constraint_violation_rate": 1.0,
100
+ "finger_joint_limit_saturation_rate": 0.5,
101
+ "joint_limit_active_step_rate": 0.024714037630782035,
102
+ "finger_joint_limit_saturation_step_rate": 0.011840349039137735,
103
+ "max_joint_limit_excess_rad": 2.2444629669189453,
104
+ "robot_constraint_diagnostics": {
105
+ "joint_limit_soft_counts": {
106
+ "l_joint4": 30,
107
+ "left_index_proximal_joint": 880,
108
+ "left_middle_distal_joint": 129,
109
+ "left_middle_proximal_joint": 190,
110
+ "left_pinky_distal_joint": 74,
111
+ "left_pinky_proximal_joint": 209,
112
+ "left_ring_distal_joint": 76,
113
+ "left_ring_proximal_joint": 140,
114
+ "left_thumb_distal_joint": 68,
115
+ "left_thumb_metacarpal_joint": 559,
116
+ "left_thumb_proximal_joint": 14,
117
+ "right_pinky_proximal_joint": 11,
118
+ "right_thumb_proximal_joint": 39
119
+ },
120
+ "joint_limit_hard_counts": {
121
+ "l_joint4": 1,
122
+ "left_index_proximal_joint": 223,
123
+ "left_middle_distal_joint": 83,
124
+ "left_middle_proximal_joint": 46,
125
+ "left_pinky_distal_joint": 45,
126
+ "left_pinky_proximal_joint": 215,
127
+ "left_ring_distal_joint": 66,
128
+ "left_ring_proximal_joint": 108,
129
+ "left_thumb_distal_joint": 4,
130
+ "left_thumb_metacarpal_joint": 6,
131
+ "left_thumb_proximal_joint": 2,
132
+ "right_middle_distal_joint": 1,
133
+ "right_middle_proximal_joint": 5,
134
+ "right_pinky_proximal_joint": 18,
135
+ "right_thumb_proximal_joint": 2
136
+ },
137
+ "joint_limit_hard_episode_events": 535,
138
+ "finger_joint_limit_saturation_counts": {
139
+ "left_index_proximal_joint": 291,
140
+ "left_middle_distal_joint": 25,
141
+ "left_middle_proximal_joint": 124,
142
+ "left_pinky_distal_joint": 29,
143
+ "left_pinky_proximal_joint": 101,
144
+ "left_thumb_distal_joint": 24,
145
+ "left_thumb_metacarpal_joint": 536,
146
+ "right_thumb_proximal_joint": 16
147
+ },
148
+ "joint_limit_max_excess_by_joint": {
149
+ "l_joint4": 0.20520567893981934,
150
+ "left_index_distal_joint": 1.429459750568185e-08,
151
+ "left_index_proximal_joint": 0.499499648809433,
152
+ "left_middle_distal_joint": 0.24086666107177734,
153
+ "left_middle_proximal_joint": 1.7787070274353027,
154
+ "left_pinky_distal_joint": 0.3008873164653778,
155
+ "left_pinky_proximal_joint": 2.2444629669189453,
156
+ "left_ring_distal_joint": 0.1571594476699829,
157
+ "left_ring_proximal_joint": 0.0928114652633667,
158
+ "left_thumb_distal_joint": 0.5271201133728027,
159
+ "left_thumb_metacarpal_joint": 1.7496693134307861,
160
+ "left_thumb_proximal_joint": 0.12862801551818848,
161
+ "right_index_distal_joint": 2.8610078195434596e-10,
162
+ "right_index_proximal_joint": 0.0016745028551667929,
163
+ "right_middle_distal_joint": 0.0052017709240317345,
164
+ "right_middle_proximal_joint": 0.005803689360618591,
165
+ "right_pinky_distal_joint": 0.0024701727088540792,
166
+ "right_pinky_proximal_joint": 0.07203339785337448,
167
+ "right_ring_distal_joint": 0.003787791356444359,
168
+ "right_ring_proximal_joint": 0.004342176951467991,
169
+ "right_thumb_proximal_joint": 0.2232069969177246
170
+ },
171
+ "coverage": {
172
+ "hard_limit_source": {
173
+ "available": 26,
174
+ "unavailable": 0,
175
+ "unknown": 0
176
+ },
177
+ "joint_role_manifest": {
178
+ "available": 26,
179
+ "unavailable": 0,
180
+ "unknown": 0
181
+ }
182
+ },
183
+ "hard_limit_sources": {
184
+ "runtime.joint_pos_limits": 26
185
+ },
186
+ "soft_limit_sources": {
187
+ "runtime.soft_joint_pos_limits": 26
188
+ },
189
+ "unknown_runtime_joint_names": [],
190
+ "baseline_checks": {
191
+ "post_home": {
192
+ "checked": 26,
193
+ "hard_violation_episodes": 0
194
+ },
195
+ "post_reset": {
196
+ "checked": 26,
197
+ "hard_violation_episodes": 0
198
+ }
199
+ }
200
+ }
201
+ },
202
+ "robot_motion": {
203
+ "joint_acceleration_rms": 14.650702636611694,
204
+ "joint_effort_rms": 25.200024015347804,
205
+ "joint_jerk_rms": 1498.9529223044779,
206
+ "joint_velocity_rms": 0.3480264956345026
207
+ },
208
+ "grasp": {
209
+ "kinematic_grasp_stability_index": 0.8528569348332496,
210
+ "mean_kinematic_grasp_stability": 0.33363492018244334,
211
+ "per_object_best_kinematic_grasp_stability": {
212
+ "obj_025_mug_1": 0.6799333758196502,
213
+ "obj_031_spoon_1": 0.77396280763783
214
+ },
215
+ "grasp_gsi_diagnostics": {
216
+ "mean_anti_slip_score": 0.6398678323428152,
217
+ "mean_duration_score": 0.643739085569897,
218
+ "mean_hold_score": 0.39132554856607993,
219
+ "mean_kinematic_stability_score": 0.08855536948239291,
220
+ "mean_legacy_gsi": 0.48160400741025633,
221
+ "mean_lift_score": 0.5113336446061553,
222
+ "mean_motion_stability_score": 0.35797995139778854
223
+ }
224
+ },
225
+ "tool": {
226
+ "tool_selection_accuracy": null,
227
+ "tool_switch_success_rate": null,
228
+ "tool_switch_total": 0,
229
+ "tool_switch_successful": 0
230
+ },
231
+ "runtime": {
232
+ "avg_steps": 2679.8846153846152,
233
+ "error_count": 0
234
+ }
235
+ }
236
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/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/67_faucet_cup_water_fill/None/native_shards/001/summary.json ADDED
@@ -0,0 +1,234 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA faucet_recovery-ee8f8b886b",
5
+ "base_seed": 106700000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 26,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.6538461538461539,
50
+ "success_rate_ci": {
51
+ "mean": 0.6538461538461539,
52
+ "ci_low": 0.46153846153846156,
53
+ "ci_high": 0.8461538461538461
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.7692307692307693
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 41.99313725490086
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.6153846153846154,
64
+ "hard_violation_rate": 0.23076923076923078,
65
+ "drop_rate": 0.23076923076923078,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 8.713204861968313e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.6538461538461539,
74
+ "at_end_success_rate": 0.6538461538461539,
75
+ "task_family_success_rate": {
76
+ "articulation_multi": 0.6538461538461539
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.5673076923076923,
81
+ "current_chain_depth_progress_score": 0.7692307692307693,
82
+ "latched_chain_depth_progress_score": 0.7692307692307693,
83
+ "avg_current_chain_depth": 3.076923076923077,
84
+ "avg_latched_chain_depth": 3.076923076923077
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7675353766902009,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2519.5882352941176,
91
+ "avg_policy_steps_to_success": 840.2352941176471,
92
+ "avg_policy_queries_to_success": 840.2352941176471
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 8.713204861968313e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_025_mug_1": 1,
98
+ "drop_to_table_obj_031_spoon_1": 5
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.46153846153846156,
102
+ "joint_limit_active_step_rate": 0.02481811184850641,
103
+ "finger_joint_limit_saturation_step_rate": 0.0074933561812927495,
104
+ "max_joint_limit_excess_rad": 1.3506115674972534,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "left_index_proximal_joint": 890,
108
+ "left_middle_distal_joint": 126,
109
+ "left_middle_proximal_joint": 109,
110
+ "left_pinky_distal_joint": 236,
111
+ "left_pinky_proximal_joint": 137,
112
+ "left_ring_distal_joint": 229,
113
+ "left_ring_proximal_joint": 143,
114
+ "left_thumb_distal_joint": 127,
115
+ "left_thumb_metacarpal_joint": 231,
116
+ "left_thumb_proximal_joint": 35,
117
+ "right_middle_proximal_joint": 1,
118
+ "right_pinky_proximal_joint": 16
119
+ },
120
+ "joint_limit_hard_counts": {
121
+ "left_index_proximal_joint": 232,
122
+ "left_middle_distal_joint": 66,
123
+ "left_middle_proximal_joint": 54,
124
+ "left_pinky_distal_joint": 106,
125
+ "left_pinky_proximal_joint": 247,
126
+ "left_ring_distal_joint": 118,
127
+ "left_ring_proximal_joint": 148,
128
+ "left_thumb_distal_joint": 10,
129
+ "left_thumb_metacarpal_joint": 4,
130
+ "left_thumb_proximal_joint": 3,
131
+ "right_middle_distal_joint": 1,
132
+ "right_middle_proximal_joint": 5,
133
+ "right_pinky_proximal_joint": 19,
134
+ "right_ring_proximal_joint": 1
135
+ },
136
+ "joint_limit_hard_episode_events": 617,
137
+ "finger_joint_limit_saturation_counts": {
138
+ "left_index_proximal_joint": 223,
139
+ "left_middle_distal_joint": 15,
140
+ "left_middle_proximal_joint": 7,
141
+ "left_pinky_distal_joint": 36,
142
+ "left_ring_distal_joint": 26,
143
+ "left_thumb_distal_joint": 51,
144
+ "left_thumb_metacarpal_joint": 220,
145
+ "left_thumb_proximal_joint": 16
146
+ },
147
+ "joint_limit_max_excess_by_joint": {
148
+ "left_index_distal_joint": 1.6116288747980434e-08,
149
+ "left_index_proximal_joint": 0.7228826880455017,
150
+ "left_middle_distal_joint": 0.33652687072753906,
151
+ "left_middle_proximal_joint": 0.3807586431503296,
152
+ "left_pinky_distal_joint": 0.4702425003051758,
153
+ "left_pinky_proximal_joint": 0.22162079811096191,
154
+ "left_ring_distal_joint": 0.40586066246032715,
155
+ "left_ring_proximal_joint": 0.17661166191101074,
156
+ "left_thumb_distal_joint": 0.4494211673736572,
157
+ "left_thumb_metacarpal_joint": 1.3506115674972534,
158
+ "left_thumb_proximal_joint": 0.3604257106781006,
159
+ "right_index_distal_joint": 2.858065450972447e-10,
160
+ "right_index_proximal_joint": 0.0015586433000862598,
161
+ "right_middle_distal_joint": 0.005025394726544619,
162
+ "right_middle_proximal_joint": 0.010845188051462173,
163
+ "right_pinky_distal_joint": 0.0025852362159639597,
164
+ "right_pinky_proximal_joint": 0.06782203167676926,
165
+ "right_ring_distal_joint": 0.003886881284415722,
166
+ "right_ring_proximal_joint": 0.008138286881148815,
167
+ "right_thumb_proximal_joint": 0.00034201363450847566
168
+ },
169
+ "coverage": {
170
+ "hard_limit_source": {
171
+ "available": 26,
172
+ "unavailable": 0,
173
+ "unknown": 0
174
+ },
175
+ "joint_role_manifest": {
176
+ "available": 26,
177
+ "unavailable": 0,
178
+ "unknown": 0
179
+ }
180
+ },
181
+ "hard_limit_sources": {
182
+ "runtime.joint_pos_limits": 26
183
+ },
184
+ "soft_limit_sources": {
185
+ "runtime.soft_joint_pos_limits": 26
186
+ },
187
+ "unknown_runtime_joint_names": [],
188
+ "baseline_checks": {
189
+ "post_home": {
190
+ "checked": 26,
191
+ "hard_violation_episodes": 0
192
+ },
193
+ "post_reset": {
194
+ "checked": 26,
195
+ "hard_violation_episodes": 0
196
+ }
197
+ }
198
+ }
199
+ },
200
+ "robot_motion": {
201
+ "joint_acceleration_rms": 16.959727944418457,
202
+ "joint_effort_rms": 28.40940869867272,
203
+ "joint_jerk_rms": 1728.6876933869687,
204
+ "joint_velocity_rms": 0.3901639048781784
205
+ },
206
+ "grasp": {
207
+ "kinematic_grasp_stability_index": 0.8929106047678539,
208
+ "mean_kinematic_grasp_stability": 0.3513336222271047,
209
+ "per_object_best_kinematic_grasp_stability": {
210
+ "obj_025_mug_1": 0.7756834864355522,
211
+ "obj_031_spoon_1": 0.8126726293506782
212
+ },
213
+ "grasp_gsi_diagnostics": {
214
+ "mean_anti_slip_score": 0.6998280534625017,
215
+ "mean_duration_score": 0.6493674291194094,
216
+ "mean_hold_score": 0.33746223079629906,
217
+ "mean_kinematic_stability_score": 0.09297985708354306,
218
+ "mean_legacy_gsi": 0.46667817288135566,
219
+ "mean_lift_score": 0.4657140641216374,
220
+ "mean_motion_stability_score": 0.3102814319676444
221
+ }
222
+ },
223
+ "tool": {
224
+ "tool_selection_accuracy": null,
225
+ "tool_switch_success_rate": null,
226
+ "tool_switch_total": 0,
227
+ "tool_switch_successful": 0
228
+ },
229
+ "runtime": {
230
+ "avg_steps": 2648.5,
231
+ "error_count": 0
232
+ }
233
+ }
234
+ }
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/None/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/67_faucet_cup_water_fill/None/summary.json ADDED
@@ -0,0 +1,241 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA faucet_recovery-ee8f8b886b",
5
+ "base_seed": 106700000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 28,
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": 2892,
16
+ "dwell_time_s": 3.0,
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": 0.42,
53
+ "ci_high": 0.7
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.72
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": 41.38273809523703
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.54,
64
+ "hard_violation_rate": 0.16,
65
+ "drop_rate": 0.16,
66
+ "high_speed_violation_rate": 0.0,
67
+ "safety_violation_step_rate": 6.008396734436375e-05
68
+ },
69
+ "robustness": {}
70
+ },
71
+ "diagnostic_metrics": {
72
+ "completion": {
73
+ "ever_instant_success_rate": 0.56,
74
+ "at_end_success_rate": 0.56,
75
+ "task_family_success_rate": {
76
+ "articulation_multi": 0.56
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.535,
81
+ "current_chain_depth_progress_score": 0.71,
82
+ "latched_chain_depth_progress_score": 0.72,
83
+ "avg_current_chain_depth": 2.84,
84
+ "avg_latched_chain_depth": 2.88
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": 0.7797473308181486,
88
+ "expert_time_s": 32.13333333333333,
89
+ "expert_time_step": 1928,
90
+ "avg_steps_to_success": 2482.964285714286,
91
+ "avg_policy_steps_to_success": 828.0357142857143,
92
+ "avg_policy_queries_to_success": 828.0357142857143
93
+ },
94
+ "safety": {
95
+ "safety_violation_events_per_step": 6.008396734436375e-05,
96
+ "violation_counts": {
97
+ "drop_to_table_obj_025_mug_1": 1,
98
+ "drop_to_table_obj_031_spoon_1": 7
99
+ },
100
+ "robot_constraint_violation_rate": 1.0,
101
+ "finger_joint_limit_saturation_rate": 0.48,
102
+ "joint_limit_active_step_rate": 0.02451425867650041,
103
+ "finger_joint_limit_saturation_step_rate": 0.0099438965954922,
104
+ "max_joint_limit_excess_rad": 2.2444629669189453,
105
+ "robot_constraint_diagnostics": {
106
+ "joint_limit_soft_counts": {
107
+ "l_joint4": 30,
108
+ "left_index_proximal_joint": 1617,
109
+ "left_middle_distal_joint": 243,
110
+ "left_middle_proximal_joint": 296,
111
+ "left_pinky_distal_joint": 307,
112
+ "left_pinky_proximal_joint": 345,
113
+ "left_ring_distal_joint": 299,
114
+ "left_ring_proximal_joint": 283,
115
+ "left_thumb_distal_joint": 195,
116
+ "left_thumb_metacarpal_joint": 790,
117
+ "left_thumb_proximal_joint": 49,
118
+ "right_middle_proximal_joint": 1,
119
+ "right_pinky_proximal_joint": 27,
120
+ "right_thumb_proximal_joint": 39
121
+ },
122
+ "joint_limit_hard_counts": {
123
+ "l_joint4": 1,
124
+ "left_index_proximal_joint": 384,
125
+ "left_middle_distal_joint": 146,
126
+ "left_middle_proximal_joint": 97,
127
+ "left_pinky_distal_joint": 146,
128
+ "left_pinky_proximal_joint": 450,
129
+ "left_ring_distal_joint": 177,
130
+ "left_ring_proximal_joint": 248,
131
+ "left_thumb_distal_joint": 14,
132
+ "left_thumb_metacarpal_joint": 10,
133
+ "left_thumb_proximal_joint": 5,
134
+ "right_middle_distal_joint": 2,
135
+ "right_middle_proximal_joint": 9,
136
+ "right_pinky_proximal_joint": 37,
137
+ "right_ring_proximal_joint": 1,
138
+ "right_thumb_proximal_joint": 2
139
+ },
140
+ "joint_limit_hard_episode_events": 1063,
141
+ "finger_joint_limit_saturation_counts": {
142
+ "left_index_proximal_joint": 497,
143
+ "left_middle_distal_joint": 40,
144
+ "left_middle_proximal_joint": 131,
145
+ "left_pinky_distal_joint": 65,
146
+ "left_pinky_proximal_joint": 101,
147
+ "left_ring_distal_joint": 26,
148
+ "left_thumb_distal_joint": 75,
149
+ "left_thumb_metacarpal_joint": 756,
150
+ "left_thumb_proximal_joint": 16,
151
+ "right_thumb_proximal_joint": 16
152
+ },
153
+ "joint_limit_max_excess_by_joint": {
154
+ "l_joint4": 0.20520567893981934,
155
+ "left_index_distal_joint": 1.6116288747980434e-08,
156
+ "left_index_proximal_joint": 0.7228826880455017,
157
+ "left_middle_distal_joint": 0.33652687072753906,
158
+ "left_middle_proximal_joint": 1.7787070274353027,
159
+ "left_pinky_distal_joint": 0.4702425003051758,
160
+ "left_pinky_proximal_joint": 2.2444629669189453,
161
+ "left_ring_distal_joint": 0.40586066246032715,
162
+ "left_ring_proximal_joint": 0.17661166191101074,
163
+ "left_thumb_distal_joint": 0.5271201133728027,
164
+ "left_thumb_metacarpal_joint": 1.7496693134307861,
165
+ "left_thumb_proximal_joint": 0.3604257106781006,
166
+ "right_index_distal_joint": 2.8610078195434596e-10,
167
+ "right_index_proximal_joint": 0.0016745028551667929,
168
+ "right_middle_distal_joint": 0.0052017709240317345,
169
+ "right_middle_proximal_joint": 0.010845188051462173,
170
+ "right_pinky_distal_joint": 0.0025852362159639597,
171
+ "right_pinky_proximal_joint": 0.07203339785337448,
172
+ "right_ring_distal_joint": 0.003886881284415722,
173
+ "right_ring_proximal_joint": 0.008138286881148815,
174
+ "right_thumb_proximal_joint": 0.2232069969177246
175
+ },
176
+ "coverage": {
177
+ "hard_limit_source": {
178
+ "available": 50,
179
+ "unavailable": 0,
180
+ "unknown": 0
181
+ },
182
+ "joint_role_manifest": {
183
+ "available": 50,
184
+ "unavailable": 0,
185
+ "unknown": 0
186
+ }
187
+ },
188
+ "hard_limit_sources": {
189
+ "runtime.joint_pos_limits": 50
190
+ },
191
+ "soft_limit_sources": {
192
+ "runtime.soft_joint_pos_limits": 50
193
+ },
194
+ "unknown_runtime_joint_names": [],
195
+ "baseline_checks": {
196
+ "post_home": {
197
+ "checked": 50,
198
+ "hard_violation_episodes": 0
199
+ },
200
+ "post_reset": {
201
+ "checked": 50,
202
+ "hard_violation_episodes": 0
203
+ }
204
+ }
205
+ }
206
+ },
207
+ "robot_motion": {
208
+ "joint_acceleration_rms": 15.692756420393296,
209
+ "joint_effort_rms": 26.928993091532337,
210
+ "joint_jerk_rms": 1601.4252136661953,
211
+ "joint_velocity_rms": 0.3680797985709752
212
+ },
213
+ "grasp": {
214
+ "kinematic_grasp_stability_index": 0.8758254845333595,
215
+ "mean_kinematic_grasp_stability": 0.34351421384509806,
216
+ "per_object_best_kinematic_grasp_stability": {
217
+ "obj_025_mug_1": 0.7312116151736756,
218
+ "obj_031_spoon_1": 0.796429606727225
219
+ },
220
+ "grasp_gsi_diagnostics": {
221
+ "mean_anti_slip_score": 0.6764612368960389,
222
+ "mean_duration_score": 0.6430645595794547,
223
+ "mean_hold_score": 0.36219665578683796,
224
+ "mean_kinematic_stability_score": 0.09144772591508508,
225
+ "mean_legacy_gsi": 0.4733436173991214,
226
+ "mean_lift_score": 0.48545670366842425,
227
+ "mean_motion_stability_score": 0.3317543523048018
228
+ }
229
+ },
230
+ "tool": {
231
+ "tool_selection_accuracy": null,
232
+ "tool_switch_success_rate": null,
233
+ "tool_switch_total": 0,
234
+ "tool_switch_successful": 0
235
+ },
236
+ "runtime": {
237
+ "avg_steps": 2662.94,
238
+ "error_count": 0
239
+ }
240
+ }
241
+ }
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/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/73_jigsaw_puzzle_assembly/Equi/native_shards/000/summary.json ADDED
@@ -0,0 +1,220 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA yellowbounds-df764f33e4",
5
+ "base_seed": 107300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 0,
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": 3639,
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": 0.0,
50
+ "success_rate_ci": {
51
+ "mean": 0.0,
52
+ "ci_low": 0.0,
53
+ "ci_high": 0.0
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.37
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": null
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.0,
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": 0.0,
74
+ "at_end_success_rate": 0.0,
75
+ "task_family_success_rate": {
76
+ "tetris_block_manipulation": 0.0
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.23,
81
+ "current_chain_depth_progress_score": 0.64,
82
+ "latched_chain_depth_progress_score": 0.96,
83
+ "avg_current_chain_depth": 0.64,
84
+ "avg_latched_chain_depth": 0.96
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": null,
88
+ "expert_time_s": 40.43333333333333,
89
+ "expert_time_step": 2426,
90
+ "avg_steps_to_success": null,
91
+ "avg_policy_steps_to_success": null,
92
+ "avg_policy_queries_to_success": null
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": 0.44,
99
+ "joint_limit_active_step_rate": 0.015059082165430063,
100
+ "finger_joint_limit_saturation_step_rate": 0.0034624896949711458,
101
+ "max_joint_limit_excess_rad": 0.4793011546134949,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_index_MCP_FE": 496,
105
+ "left_index_PIP": 880,
106
+ "multi_right_index_DIP": 15,
107
+ "multi_right_index_PIP": 14
108
+ },
109
+ "joint_limit_hard_counts": {
110
+ "left_index_MCP_FE": 20,
111
+ "left_index_PIP": 484,
112
+ "left_pinky_CMC": 12,
113
+ "multi_right_index_DIP": 1,
114
+ "multi_right_index_PIP": 9
115
+ },
116
+ "joint_limit_hard_episode_events": 476,
117
+ "finger_joint_limit_saturation_counts": {
118
+ "left_index_MCP_FE": 195,
119
+ "left_index_PIP": 110,
120
+ "multi_right_index_DIP": 10
121
+ },
122
+ "joint_limit_max_excess_by_joint": {
123
+ "A7": 0.0004162788391113281,
124
+ "left_index_DIP": 1.7361224308842793e-05,
125
+ "left_index_MCP_FE": 0.3346407413482666,
126
+ "left_index_PIP": 0.24967479705810547,
127
+ "left_middle_DIP": 0.0001920461654663086,
128
+ "left_middle_MCP_FE": 0.0004514455795288086,
129
+ "left_middle_PIP": 0.0019683837890625,
130
+ "left_pinky_CMC": 0.008165400475263596,
131
+ "left_pinky_DIP": 0.0005167722702026367,
132
+ "left_pinky_MCP_FE": 0.00015211105346679688,
133
+ "left_pinky_PIP": 0.0012242794036865234,
134
+ "left_ring_DIP": 0.0003198385238647461,
135
+ "left_ring_MCP_FE": 7.343292236328125e-05,
136
+ "left_ring_PIP": 0.0016480684280395508,
137
+ "left_thumb_IP": 6.180844724923418e-11,
138
+ "multi_right_index_DIP": 0.4793011546134949,
139
+ "multi_right_index_PIP": 0.17510807514190674,
140
+ "multi_right_middle_DIP": 6.508827209472656e-05,
141
+ "multi_right_middle_MCP_FE": 0.00020325183868408203,
142
+ "multi_right_middle_PIP": 0.0015282630920410156,
143
+ "multi_right_pinky_CMC": 0.0032918236684054136,
144
+ "multi_right_pinky_DIP": 0.00013935565948486328,
145
+ "multi_right_pinky_MCP_FE": 0.0007183551788330078,
146
+ "multi_right_pinky_PIP": 0.0009119510650634766,
147
+ "multi_right_ring_DIP": 8.45193862915039e-05,
148
+ "multi_right_ring_MCP_FE": 0.00022971630096435547,
149
+ "multi_right_ring_PIP": 0.0010460615158081055,
150
+ "multi_right_thumb_IP": 1.955746120607671e-10
151
+ },
152
+ "coverage": {
153
+ "hard_limit_source": {
154
+ "available": 25,
155
+ "unavailable": 0,
156
+ "unknown": 0
157
+ },
158
+ "joint_role_manifest": {
159
+ "available": 25,
160
+ "unavailable": 0,
161
+ "unknown": 0
162
+ }
163
+ },
164
+ "hard_limit_sources": {
165
+ "runtime.joint_pos_limits": 25
166
+ },
167
+ "soft_limit_sources": {
168
+ "runtime.soft_joint_pos_limits": 25
169
+ },
170
+ "unknown_runtime_joint_names": [],
171
+ "baseline_checks": {
172
+ "post_home": {
173
+ "checked": 25,
174
+ "hard_violation_episodes": 0
175
+ },
176
+ "post_reset": {
177
+ "checked": 25,
178
+ "hard_violation_episodes": 0
179
+ }
180
+ }
181
+ }
182
+ },
183
+ "robot_motion": {
184
+ "joint_acceleration_rms": 17.323950031051634,
185
+ "joint_effort_rms": 31.809591171462266,
186
+ "joint_jerk_rms": 1792.0968154681207,
187
+ "joint_velocity_rms": 0.5191229743295578
188
+ },
189
+ "grasp": {
190
+ "kinematic_grasp_stability_index": 0.012014597348491482,
191
+ "mean_kinematic_grasp_stability": 0.05745536904546628,
192
+ "per_object_best_kinematic_grasp_stability": {
193
+ "obj_326_piece_1": 0.0,
194
+ "obj_326_piece_2": 0.0,
195
+ "obj_326_piece_3": 0.012014597348491482,
196
+ "obj_326_piece_4": 0.0,
197
+ "obj_326_piece_5": 0.0
198
+ },
199
+ "grasp_gsi_diagnostics": {
200
+ "mean_anti_slip_score": 1.0,
201
+ "mean_duration_score": 0.0,
202
+ "mean_hold_score": 1.3410277548777134e-05,
203
+ "mean_kinematic_stability_score": 3.6011468809441033e-06,
204
+ "mean_legacy_gsi": 0.20001542391591348,
205
+ "mean_lift_score": 8.353943391041495e-05,
206
+ "mean_motion_stability_score": 2.006002117790006e-05
207
+ }
208
+ },
209
+ "tool": {
210
+ "tool_selection_accuracy": null,
211
+ "tool_switch_success_rate": null,
212
+ "tool_switch_total": 0,
213
+ "tool_switch_successful": 0
214
+ },
215
+ "runtime": {
216
+ "avg_steps": 3639,
217
+ "error_count": 0
218
+ }
219
+ }
220
+ }
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/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/73_jigsaw_puzzle_assembly/Equi/native_shards/001/summary.json ADDED
@@ -0,0 +1,218 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA yellowbounds-df764f33e4",
5
+ "base_seed": 107300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 0,
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": 3639,
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": 0.0,
50
+ "success_rate_ci": {
51
+ "mean": 0.0,
52
+ "ci_low": 0.0,
53
+ "ci_high": 0.0
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.37
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": null
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.0,
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": 0.0,
74
+ "at_end_success_rate": 0.0,
75
+ "task_family_success_rate": {
76
+ "tetris_block_manipulation": 0.0
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.18,
81
+ "current_chain_depth_progress_score": 0.52,
82
+ "latched_chain_depth_progress_score": 0.92,
83
+ "avg_current_chain_depth": 0.52,
84
+ "avg_latched_chain_depth": 0.92
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": null,
88
+ "expert_time_s": 40.43333333333333,
89
+ "expert_time_step": 2426,
90
+ "avg_steps_to_success": null,
91
+ "avg_policy_steps_to_success": null,
92
+ "avg_policy_queries_to_success": null
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": 0.6,
99
+ "joint_limit_active_step_rate": 0.014179719703215169,
100
+ "finger_joint_limit_saturation_step_rate": 0.0029128881560868372,
101
+ "max_joint_limit_excess_rad": 0.3472200632095337,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_index_MCP_FE": 502,
105
+ "left_index_PIP": 795,
106
+ "multi_right_index_PIP": 23
107
+ },
108
+ "joint_limit_hard_counts": {
109
+ "left_index_MCP_FE": 18,
110
+ "left_index_PIP": 427,
111
+ "left_pinky_CMC": 7,
112
+ "multi_right_index_PIP": 15
113
+ },
114
+ "joint_limit_hard_episode_events": 399,
115
+ "finger_joint_limit_saturation_counts": {
116
+ "left_index_MCP_FE": 235,
117
+ "left_index_PIP": 29,
118
+ "multi_right_index_PIP": 1
119
+ },
120
+ "joint_limit_max_excess_by_joint": {
121
+ "A7": 0.0004038810729980469,
122
+ "left_index_DIP": 1.366826836601831e-05,
123
+ "left_index_MCP_FE": 0.3472200632095337,
124
+ "left_index_PIP": 0.32424604892730713,
125
+ "left_middle_DIP": 0.00013327598571777344,
126
+ "left_middle_MCP_FE": 0.00021910667419433594,
127
+ "left_middle_PIP": 0.0027071237564086914,
128
+ "left_pinky_CMC": 0.007593141868710518,
129
+ "left_pinky_DIP": 0.0002669095993041992,
130
+ "left_pinky_MCP_FE": 0.00019598007202148438,
131
+ "left_pinky_PIP": 0.0014393329620361328,
132
+ "left_ring_DIP": 0.00014543533325195312,
133
+ "left_ring_MCP_FE": 0.00010454654693603516,
134
+ "left_ring_PIP": 0.0015583038330078125,
135
+ "left_thumb_IP": 8.705619558568856e-11,
136
+ "multi_right_index_DIP": 1.3060720448265783e-05,
137
+ "multi_right_index_PIP": 0.2570744752883911,
138
+ "multi_right_middle_DIP": 8.96453857421875e-05,
139
+ "multi_right_middle_MCP_FE": 0.00034415721893310547,
140
+ "multi_right_middle_PIP": 0.0014543533325195312,
141
+ "multi_right_pinky_CMC": 0.002796826185658574,
142
+ "multi_right_pinky_DIP": 0.0004677772521972656,
143
+ "multi_right_pinky_MCP_FE": 0.0006048679351806641,
144
+ "multi_right_pinky_PIP": 0.00032341480255126953,
145
+ "multi_right_ring_DIP": 0.00025069713592529297,
146
+ "multi_right_ring_MCP_FE": 0.00018608570098876953,
147
+ "multi_right_ring_PIP": 0.0004156827926635742,
148
+ "multi_right_thumb_IP": 1.027713691925669e-09
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": 17.2837872584428,
183
+ "joint_effort_rms": 31.398865662123715,
184
+ "joint_jerk_rms": 1788.2664958094406,
185
+ "joint_velocity_rms": 0.5208587250164033
186
+ },
187
+ "grasp": {
188
+ "kinematic_grasp_stability_index": 0.14357301978952552,
189
+ "mean_kinematic_grasp_stability": 0.2218885129011641,
190
+ "per_object_best_kinematic_grasp_stability": {
191
+ "obj_326_piece_1": 0.0,
192
+ "obj_326_piece_2": 0.070646540313712,
193
+ "obj_326_piece_3": 0.0711979473693461,
194
+ "obj_326_piece_4": 0.0,
195
+ "obj_326_piece_5": 0.06626876164719976
196
+ },
197
+ "grasp_gsi_diagnostics": {
198
+ "mean_anti_slip_score": 0.9990546853531185,
199
+ "mean_duration_score": 0.004199468718512393,
200
+ "mean_hold_score": 0.0017171750480901343,
201
+ "mean_kinematic_stability_score": 0.0015851452783351126,
202
+ "mean_legacy_gsi": 0.20172506854196573,
203
+ "mean_lift_score": 0.002268755152514427,
204
+ "mean_motion_stability_score": 0.0017533390495065625
205
+ }
206
+ },
207
+ "tool": {
208
+ "tool_selection_accuracy": null,
209
+ "tool_switch_success_rate": null,
210
+ "tool_switch_total": 0,
211
+ "tool_switch_successful": 0
212
+ },
213
+ "runtime": {
214
+ "avg_steps": 3639,
215
+ "error_count": 0
216
+ }
217
+ }
218
+ }
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Equi/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Equi/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 yellowbounds-df764f33e4",
5
+ "base_seed": 107300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 0,
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": 3639,
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": 0.0,
50
+ "success_rate_ci": {
51
+ "mean": 0.0,
52
+ "ci_low": 0.0,
53
+ "ci_high": 0.0
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.37
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": null
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.0,
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": 0.0,
74
+ "at_end_success_rate": 0.0,
75
+ "task_family_success_rate": {
76
+ "tetris_block_manipulation": 0.0
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.205,
81
+ "current_chain_depth_progress_score": 0.58,
82
+ "latched_chain_depth_progress_score": 0.94,
83
+ "avg_current_chain_depth": 0.58,
84
+ "avg_latched_chain_depth": 0.94
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": null,
88
+ "expert_time_s": 40.43333333333333,
89
+ "expert_time_step": 2426,
90
+ "avg_steps_to_success": null,
91
+ "avg_policy_steps_to_success": null,
92
+ "avg_policy_queries_to_success": null
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": 0.52,
99
+ "joint_limit_active_step_rate": 0.014619400934322617,
100
+ "finger_joint_limit_saturation_step_rate": 0.0031876889255289913,
101
+ "max_joint_limit_excess_rad": 0.4793011546134949,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_index_MCP_FE": 998,
105
+ "left_index_PIP": 1675,
106
+ "multi_right_index_DIP": 15,
107
+ "multi_right_index_PIP": 37
108
+ },
109
+ "joint_limit_hard_counts": {
110
+ "left_index_MCP_FE": 38,
111
+ "left_index_PIP": 911,
112
+ "left_pinky_CMC": 19,
113
+ "multi_right_index_DIP": 1,
114
+ "multi_right_index_PIP": 24
115
+ },
116
+ "joint_limit_hard_episode_events": 875,
117
+ "finger_joint_limit_saturation_counts": {
118
+ "left_index_MCP_FE": 430,
119
+ "left_index_PIP": 139,
120
+ "multi_right_index_DIP": 10,
121
+ "multi_right_index_PIP": 1
122
+ },
123
+ "joint_limit_max_excess_by_joint": {
124
+ "A7": 0.0004162788391113281,
125
+ "left_index_DIP": 1.7361224308842793e-05,
126
+ "left_index_MCP_FE": 0.3472200632095337,
127
+ "left_index_PIP": 0.32424604892730713,
128
+ "left_middle_DIP": 0.0001920461654663086,
129
+ "left_middle_MCP_FE": 0.0004514455795288086,
130
+ "left_middle_PIP": 0.0027071237564086914,
131
+ "left_pinky_CMC": 0.008165400475263596,
132
+ "left_pinky_DIP": 0.0005167722702026367,
133
+ "left_pinky_MCP_FE": 0.00019598007202148438,
134
+ "left_pinky_PIP": 0.0014393329620361328,
135
+ "left_ring_DIP": 0.0003198385238647461,
136
+ "left_ring_MCP_FE": 0.00010454654693603516,
137
+ "left_ring_PIP": 0.0016480684280395508,
138
+ "left_thumb_IP": 8.705619558568856e-11,
139
+ "multi_right_index_DIP": 0.4793011546134949,
140
+ "multi_right_index_PIP": 0.2570744752883911,
141
+ "multi_right_middle_DIP": 8.96453857421875e-05,
142
+ "multi_right_middle_MCP_FE": 0.00034415721893310547,
143
+ "multi_right_middle_PIP": 0.0015282630920410156,
144
+ "multi_right_pinky_CMC": 0.0032918236684054136,
145
+ "multi_right_pinky_DIP": 0.0004677772521972656,
146
+ "multi_right_pinky_MCP_FE": 0.0007183551788330078,
147
+ "multi_right_pinky_PIP": 0.0009119510650634766,
148
+ "multi_right_ring_DIP": 0.00025069713592529297,
149
+ "multi_right_ring_MCP_FE": 0.00022971630096435547,
150
+ "multi_right_ring_PIP": 0.0010460615158081055,
151
+ "multi_right_thumb_IP": 1.027713691925669e-09
152
+ },
153
+ "coverage": {
154
+ "hard_limit_source": {
155
+ "available": 50,
156
+ "unavailable": 0,
157
+ "unknown": 0
158
+ },
159
+ "joint_role_manifest": {
160
+ "available": 50,
161
+ "unavailable": 0,
162
+ "unknown": 0
163
+ }
164
+ },
165
+ "hard_limit_sources": {
166
+ "runtime.joint_pos_limits": 50
167
+ },
168
+ "soft_limit_sources": {
169
+ "runtime.soft_joint_pos_limits": 50
170
+ },
171
+ "unknown_runtime_joint_names": [],
172
+ "baseline_checks": {
173
+ "post_home": {
174
+ "checked": 50,
175
+ "hard_violation_episodes": 0
176
+ },
177
+ "post_reset": {
178
+ "checked": 50,
179
+ "hard_violation_episodes": 0
180
+ }
181
+ }
182
+ }
183
+ },
184
+ "robot_motion": {
185
+ "joint_acceleration_rms": 17.30386864474722,
186
+ "joint_effort_rms": 31.60422841679299,
187
+ "joint_jerk_rms": 1790.1816556387807,
188
+ "joint_velocity_rms": 0.5199908496729805
189
+ },
190
+ "grasp": {
191
+ "kinematic_grasp_stability_index": 0.07779380856900851,
192
+ "mean_kinematic_grasp_stability": 0.18078022693723964,
193
+ "per_object_best_kinematic_grasp_stability": {
194
+ "obj_326_piece_1": 0.0,
195
+ "obj_326_piece_2": 0.035323270156856,
196
+ "obj_326_piece_3": 0.0416062723589188,
197
+ "obj_326_piece_4": 0.0,
198
+ "obj_326_piece_5": 0.03313438082359988
199
+ },
200
+ "grasp_gsi_diagnostics": {
201
+ "mean_anti_slip_score": 0.9995273426765592,
202
+ "mean_duration_score": 0.0020997343592561966,
203
+ "mean_hold_score": 0.0008652926628194558,
204
+ "mean_kinematic_stability_score": 0.0007943732126080283,
205
+ "mean_legacy_gsi": 0.2008702462289396,
206
+ "mean_lift_score": 0.001176147293212421,
207
+ "mean_motion_stability_score": 0.0008866995353422313
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": 3639,
218
+ "error_count": 0
219
+ }
220
+ }
221
+ }
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/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/73_jigsaw_puzzle_assembly/Full/native_shards/000/summary.json ADDED
@@ -0,0 +1,217 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA yellowbounds-df764f33e4",
5
+ "base_seed": 207300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 0,
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": 3639,
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": 0.0,
50
+ "success_rate_ci": {
51
+ "mean": 0.0,
52
+ "ci_low": 0.0,
53
+ "ci_high": 0.0
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.38
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": null
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.0,
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": 0.0,
74
+ "at_end_success_rate": 0.0,
75
+ "task_family_success_rate": {
76
+ "tetris_block_manipulation": 0.0
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.31,
81
+ "current_chain_depth_progress_score": 0.8,
82
+ "latched_chain_depth_progress_score": 0.92,
83
+ "avg_current_chain_depth": 0.8,
84
+ "avg_latched_chain_depth": 0.92
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": null,
88
+ "expert_time_s": 40.43333333333333,
89
+ "expert_time_step": 2426,
90
+ "avg_steps_to_success": null,
91
+ "avg_policy_steps_to_success": null,
92
+ "avg_policy_queries_to_success": null
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": 0.56,
99
+ "joint_limit_active_step_rate": 0.017345424567188787,
100
+ "finger_joint_limit_saturation_step_rate": 0.003781258587524045,
101
+ "max_joint_limit_excess_rad": 0.41419970989227295,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_index_MCP_FE": 605,
105
+ "left_index_PIP": 1016,
106
+ "multi_right_index_PIP": 20
107
+ },
108
+ "joint_limit_hard_counts": {
109
+ "left_index_MCP_FE": 20,
110
+ "left_index_PIP": 478,
111
+ "left_pinky_CMC": 3,
112
+ "multi_right_index_PIP": 20
113
+ },
114
+ "joint_limit_hard_episode_events": 454,
115
+ "finger_joint_limit_saturation_counts": {
116
+ "left_index_MCP_FE": 272,
117
+ "left_index_PIP": 72
118
+ },
119
+ "joint_limit_max_excess_by_joint": {
120
+ "A7": 0.0003058910369873047,
121
+ "left_index_DIP": 1.7103715435951017e-05,
122
+ "left_index_MCP_FE": 0.41419970989227295,
123
+ "left_index_PIP": 0.32698965072631836,
124
+ "left_middle_DIP": 0.0001214742660522461,
125
+ "left_middle_MCP_FE": 0.00024700164794921875,
126
+ "left_middle_PIP": 0.002776503562927246,
127
+ "left_pinky_CMC": 0.005543834529817104,
128
+ "left_pinky_DIP": 0.00028073787689208984,
129
+ "left_pinky_MCP_FE": 0.0001252889633178711,
130
+ "left_pinky_PIP": 0.0013115406036376953,
131
+ "left_ring_DIP": 0.00015532970428466797,
132
+ "left_ring_MCP_FE": 9.107589721679688e-05,
133
+ "left_ring_PIP": 0.0012873411178588867,
134
+ "left_thumb_IP": 6.110515565760366e-11,
135
+ "multi_right_index_DIP": 3.3288069971604273e-06,
136
+ "multi_right_index_PIP": 0.09314948320388794,
137
+ "multi_right_middle_DIP": 6.973743438720703e-05,
138
+ "multi_right_middle_MCP_FE": 0.00019276142120361328,
139
+ "multi_right_middle_PIP": 0.001688838005065918,
140
+ "multi_right_pinky_CMC": 0.002621823688969016,
141
+ "multi_right_pinky_DIP": 0.00019741058349609375,
142
+ "multi_right_pinky_MCP_FE": 0.0005128383636474609,
143
+ "multi_right_pinky_PIP": 0.0005137920379638672,
144
+ "multi_right_ring_DIP": 0.00011301040649414062,
145
+ "multi_right_ring_MCP_FE": 0.00017178058624267578,
146
+ "multi_right_ring_PIP": 0.0007424354553222656,
147
+ "multi_right_thumb_IP": 1.0177018117119019e-09
148
+ },
149
+ "coverage": {
150
+ "hard_limit_source": {
151
+ "available": 25,
152
+ "unavailable": 0,
153
+ "unknown": 0
154
+ },
155
+ "joint_role_manifest": {
156
+ "available": 25,
157
+ "unavailable": 0,
158
+ "unknown": 0
159
+ }
160
+ },
161
+ "hard_limit_sources": {
162
+ "runtime.joint_pos_limits": 25
163
+ },
164
+ "soft_limit_sources": {
165
+ "runtime.soft_joint_pos_limits": 25
166
+ },
167
+ "unknown_runtime_joint_names": [],
168
+ "baseline_checks": {
169
+ "post_home": {
170
+ "checked": 25,
171
+ "hard_violation_episodes": 0
172
+ },
173
+ "post_reset": {
174
+ "checked": 25,
175
+ "hard_violation_episodes": 0
176
+ }
177
+ }
178
+ }
179
+ },
180
+ "robot_motion": {
181
+ "joint_acceleration_rms": 17.226137063355367,
182
+ "joint_effort_rms": 31.77486636090701,
183
+ "joint_jerk_rms": 1783.336128471949,
184
+ "joint_velocity_rms": 0.5177571381872549
185
+ },
186
+ "grasp": {
187
+ "kinematic_grasp_stability_index": 0.07975180854749742,
188
+ "mean_kinematic_grasp_stability": 0.12979400320284973,
189
+ "per_object_best_kinematic_grasp_stability": {
190
+ "obj_326_piece_1": 0.0,
191
+ "obj_326_piece_2": 0.002873941197968437,
192
+ "obj_326_piece_3": 0.07687786734952899,
193
+ "obj_326_piece_4": 0.0,
194
+ "obj_326_piece_5": 0.0
195
+ },
196
+ "grasp_gsi_diagnostics": {
197
+ "mean_anti_slip_score": 1.0,
198
+ "mean_duration_score": 0.0007531739488870593,
199
+ "mean_hold_score": 3.7372904644133e-05,
200
+ "mean_kinematic_stability_score": 1.7778109899227822e-05,
201
+ "mean_legacy_gsi": 0.20018301195169153,
202
+ "mean_lift_score": 0.00013190436933223414,
203
+ "mean_motion_stability_score": 5.618253042927461e-05
204
+ }
205
+ },
206
+ "tool": {
207
+ "tool_selection_accuracy": null,
208
+ "tool_switch_success_rate": null,
209
+ "tool_switch_total": 0,
210
+ "tool_switch_successful": 0
211
+ },
212
+ "runtime": {
213
+ "avg_steps": 3639,
214
+ "error_count": 0
215
+ }
216
+ }
217
+ }
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/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/73_jigsaw_puzzle_assembly/Full/native_shards/001/summary.json ADDED
@@ -0,0 +1,218 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA yellowbounds-df764f33e4",
5
+ "base_seed": 207300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 0,
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": 3639,
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": 0.0,
50
+ "success_rate_ci": {
51
+ "mean": 0.0,
52
+ "ci_low": 0.0,
53
+ "ci_high": 0.0
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.36
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": null
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.0,
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": 0.0,
74
+ "at_end_success_rate": 0.0,
75
+ "task_family_success_rate": {
76
+ "tetris_block_manipulation": 0.0
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.24,
81
+ "current_chain_depth_progress_score": 0.76,
82
+ "latched_chain_depth_progress_score": 0.88,
83
+ "avg_current_chain_depth": 0.76,
84
+ "avg_latched_chain_depth": 0.88
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": null,
88
+ "expert_time_s": 40.43333333333333,
89
+ "expert_time_step": 2426,
90
+ "avg_steps_to_success": null,
91
+ "avg_policy_steps_to_success": null,
92
+ "avg_policy_queries_to_success": null
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": 0.4,
99
+ "joint_limit_active_step_rate": 0.01348722176422094,
100
+ "finger_joint_limit_saturation_step_rate": 0.003143720802418247,
101
+ "max_joint_limit_excess_rad": 0.3935309648513794,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_index_MCP_FE": 617,
105
+ "left_index_PIP": 656,
106
+ "left_pinky_CMC": 1,
107
+ "multi_right_index_PIP": 16
108
+ },
109
+ "joint_limit_hard_counts": {
110
+ "left_index_MCP_FE": 22,
111
+ "left_index_PIP": 495,
112
+ "left_pinky_CMC": 14,
113
+ "multi_right_index_PIP": 10
114
+ },
115
+ "joint_limit_hard_episode_events": 459,
116
+ "finger_joint_limit_saturation_counts": {
117
+ "left_index_MCP_FE": 263,
118
+ "left_index_PIP": 23
119
+ },
120
+ "joint_limit_max_excess_by_joint": {
121
+ "A7": 0.0004057884216308594,
122
+ "left_index_DIP": 1.62772284966195e-05,
123
+ "left_index_MCP_FE": 0.3935309648513794,
124
+ "left_index_PIP": 0.22776944935321808,
125
+ "left_middle_DIP": 0.00018346309661865234,
126
+ "left_middle_MCP_FE": 0.0003732442855834961,
127
+ "left_middle_PIP": 0.0018820762634277344,
128
+ "left_pinky_CMC": 0.014959685504436493,
129
+ "left_pinky_DIP": 0.00029659271240234375,
130
+ "left_pinky_MCP_FE": 0.0004515647888183594,
131
+ "left_pinky_PIP": 0.0014989376068115234,
132
+ "left_ring_DIP": 0.00018513202667236328,
133
+ "left_ring_MCP_FE": 0.0002295970916748047,
134
+ "left_ring_PIP": 0.0011767148971557617,
135
+ "left_thumb_IP": 2.651788510998898e-10,
136
+ "multi_right_index_DIP": 1.2231667824380565e-05,
137
+ "multi_right_index_PIP": 0.14030008018016815,
138
+ "multi_right_middle_DIP": 0.00012922286987304688,
139
+ "multi_right_middle_MCP_FE": 0.0003281831741333008,
140
+ "multi_right_middle_PIP": 0.0016222000122070312,
141
+ "multi_right_pinky_CMC": 0.0036113352980464697,
142
+ "multi_right_pinky_DIP": 0.0003535747528076172,
143
+ "multi_right_pinky_MCP_FE": 0.0008927583694458008,
144
+ "multi_right_pinky_PIP": 0.0005918741226196289,
145
+ "multi_right_ring_DIP": 0.00017070770263671875,
146
+ "multi_right_ring_MCP_FE": 0.0003085136413574219,
147
+ "multi_right_ring_PIP": 0.0005091428756713867,
148
+ "multi_right_thumb_IP": 1.8973538018496328e-10
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": 17.453481777515854,
183
+ "joint_effort_rms": 31.36536449645672,
184
+ "joint_jerk_rms": 1810.4565030145836,
185
+ "joint_velocity_rms": 0.5196703530475183
186
+ },
187
+ "grasp": {
188
+ "kinematic_grasp_stability_index": 0.034224176959445074,
189
+ "mean_kinematic_grasp_stability": 0.06215358001827286,
190
+ "per_object_best_kinematic_grasp_stability": {
191
+ "obj_326_piece_1": 0.0,
192
+ "obj_326_piece_2": 0.027973802345937843,
193
+ "obj_326_piece_3": 0.0062503746135072315,
194
+ "obj_326_piece_4": 0.0,
195
+ "obj_326_piece_5": 0.0
196
+ },
197
+ "grasp_gsi_diagnostics": {
198
+ "mean_anti_slip_score": 1.0,
199
+ "mean_duration_score": 0.0006582028029678297,
200
+ "mean_hold_score": 4.671613080516626e-05,
201
+ "mean_kinematic_stability_score": 2.0028201466296698e-05,
202
+ "mean_legacy_gsi": 0.2001752116990788,
203
+ "mean_lift_score": 0.00019565814784281395,
204
+ "mean_motion_stability_score": 6.495733559716061e-05
205
+ }
206
+ },
207
+ "tool": {
208
+ "tool_selection_accuracy": null,
209
+ "tool_switch_success_rate": null,
210
+ "tool_switch_total": 0,
211
+ "tool_switch_successful": 0
212
+ },
213
+ "runtime": {
214
+ "avg_steps": 3639,
215
+ "error_count": 0
216
+ }
217
+ }
218
+ }
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Full/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Full/summary.json ADDED
@@ -0,0 +1,218 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA yellowbounds-df764f33e4",
5
+ "base_seed": 207300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 0,
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": 3639,
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": 0.0,
50
+ "success_rate_ci": {
51
+ "mean": 0.0,
52
+ "ci_low": 0.0,
53
+ "ci_high": 0.0
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.37
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": null
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.0,
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": 0.0,
74
+ "at_end_success_rate": 0.0,
75
+ "task_family_success_rate": {
76
+ "tetris_block_manipulation": 0.0
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.275,
81
+ "current_chain_depth_progress_score": 0.78,
82
+ "latched_chain_depth_progress_score": 0.9,
83
+ "avg_current_chain_depth": 0.78,
84
+ "avg_latched_chain_depth": 0.9
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": null,
88
+ "expert_time_s": 40.43333333333333,
89
+ "expert_time_step": 2426,
90
+ "avg_steps_to_success": null,
91
+ "avg_policy_steps_to_success": null,
92
+ "avg_policy_queries_to_success": null
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": 0.48,
99
+ "joint_limit_active_step_rate": 0.015416323165704864,
100
+ "finger_joint_limit_saturation_step_rate": 0.0034624896949711458,
101
+ "max_joint_limit_excess_rad": 0.41419970989227295,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_index_MCP_FE": 1222,
105
+ "left_index_PIP": 1672,
106
+ "left_pinky_CMC": 1,
107
+ "multi_right_index_PIP": 36
108
+ },
109
+ "joint_limit_hard_counts": {
110
+ "left_index_MCP_FE": 42,
111
+ "left_index_PIP": 973,
112
+ "left_pinky_CMC": 17,
113
+ "multi_right_index_PIP": 30
114
+ },
115
+ "joint_limit_hard_episode_events": 913,
116
+ "finger_joint_limit_saturation_counts": {
117
+ "left_index_MCP_FE": 535,
118
+ "left_index_PIP": 95
119
+ },
120
+ "joint_limit_max_excess_by_joint": {
121
+ "A7": 0.0004057884216308594,
122
+ "left_index_DIP": 1.7103715435951017e-05,
123
+ "left_index_MCP_FE": 0.41419970989227295,
124
+ "left_index_PIP": 0.32698965072631836,
125
+ "left_middle_DIP": 0.00018346309661865234,
126
+ "left_middle_MCP_FE": 0.0003732442855834961,
127
+ "left_middle_PIP": 0.002776503562927246,
128
+ "left_pinky_CMC": 0.014959685504436493,
129
+ "left_pinky_DIP": 0.00029659271240234375,
130
+ "left_pinky_MCP_FE": 0.0004515647888183594,
131
+ "left_pinky_PIP": 0.0014989376068115234,
132
+ "left_ring_DIP": 0.00018513202667236328,
133
+ "left_ring_MCP_FE": 0.0002295970916748047,
134
+ "left_ring_PIP": 0.0012873411178588867,
135
+ "left_thumb_IP": 2.651788510998898e-10,
136
+ "multi_right_index_DIP": 1.2231667824380565e-05,
137
+ "multi_right_index_PIP": 0.14030008018016815,
138
+ "multi_right_middle_DIP": 0.00012922286987304688,
139
+ "multi_right_middle_MCP_FE": 0.0003281831741333008,
140
+ "multi_right_middle_PIP": 0.001688838005065918,
141
+ "multi_right_pinky_CMC": 0.0036113352980464697,
142
+ "multi_right_pinky_DIP": 0.0003535747528076172,
143
+ "multi_right_pinky_MCP_FE": 0.0008927583694458008,
144
+ "multi_right_pinky_PIP": 0.0005918741226196289,
145
+ "multi_right_ring_DIP": 0.00017070770263671875,
146
+ "multi_right_ring_MCP_FE": 0.0003085136413574219,
147
+ "multi_right_ring_PIP": 0.0007424354553222656,
148
+ "multi_right_thumb_IP": 1.0177018117119019e-09
149
+ },
150
+ "coverage": {
151
+ "hard_limit_source": {
152
+ "available": 50,
153
+ "unavailable": 0,
154
+ "unknown": 0
155
+ },
156
+ "joint_role_manifest": {
157
+ "available": 50,
158
+ "unavailable": 0,
159
+ "unknown": 0
160
+ }
161
+ },
162
+ "hard_limit_sources": {
163
+ "runtime.joint_pos_limits": 50
164
+ },
165
+ "soft_limit_sources": {
166
+ "runtime.soft_joint_pos_limits": 50
167
+ },
168
+ "unknown_runtime_joint_names": [],
169
+ "baseline_checks": {
170
+ "post_home": {
171
+ "checked": 50,
172
+ "hard_violation_episodes": 0
173
+ },
174
+ "post_reset": {
175
+ "checked": 50,
176
+ "hard_violation_episodes": 0
177
+ }
178
+ }
179
+ }
180
+ },
181
+ "robot_motion": {
182
+ "joint_acceleration_rms": 17.33980942043561,
183
+ "joint_effort_rms": 31.570115428681863,
184
+ "joint_jerk_rms": 1796.8963157432663,
185
+ "joint_velocity_rms": 0.5187137456173866
186
+ },
187
+ "grasp": {
188
+ "kinematic_grasp_stability_index": 0.05698799275347125,
189
+ "mean_kinematic_grasp_stability": 0.10724719547465744,
190
+ "per_object_best_kinematic_grasp_stability": {
191
+ "obj_326_piece_1": 0.0,
192
+ "obj_326_piece_2": 0.01542387177195314,
193
+ "obj_326_piece_3": 0.041564120981518106,
194
+ "obj_326_piece_4": 0.0,
195
+ "obj_326_piece_5": 0.0
196
+ },
197
+ "grasp_gsi_diagnostics": {
198
+ "mean_anti_slip_score": 1.0,
199
+ "mean_duration_score": 0.0007056883759274444,
200
+ "mean_hold_score": 4.204451772464963e-05,
201
+ "mean_kinematic_stability_score": 1.890315568276226e-05,
202
+ "mean_legacy_gsi": 0.20017911182538514,
203
+ "mean_lift_score": 0.00016378125858752406,
204
+ "mean_motion_stability_score": 6.056993301321761e-05
205
+ }
206
+ },
207
+ "tool": {
208
+ "tool_selection_accuracy": null,
209
+ "tool_switch_success_rate": null,
210
+ "tool_switch_total": 0,
211
+ "tool_switch_successful": 0
212
+ },
213
+ "runtime": {
214
+ "avg_steps": 3639,
215
+ "error_count": 0
216
+ }
217
+ }
218
+ }
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/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/73_jigsaw_puzzle_assembly/Inv/native_shards/000/summary.json ADDED
@@ -0,0 +1,217 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA yellowbounds-df764f33e4",
5
+ "base_seed": 107300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 26,
9
+ "successes": 0,
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": 3639,
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": 0.0,
50
+ "success_rate_ci": {
51
+ "mean": 0.0,
52
+ "ci_low": 0.0,
53
+ "ci_high": 0.0
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.3942307692307692
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": null
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.0,
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": 0.0,
74
+ "at_end_success_rate": 0.0,
75
+ "task_family_success_rate": {
76
+ "tetris_block_manipulation": 0.0
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.27884615384615385,
81
+ "current_chain_depth_progress_score": 0.7692307692307693,
82
+ "latched_chain_depth_progress_score": 0.9615384615384616,
83
+ "avg_current_chain_depth": 0.7692307692307693,
84
+ "avg_latched_chain_depth": 0.9615384615384616
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": null,
88
+ "expert_time_s": 40.43333333333333,
89
+ "expert_time_step": 2426,
90
+ "avg_steps_to_success": null,
91
+ "avg_policy_steps_to_success": null,
92
+ "avg_policy_queries_to_success": null
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": 0.5,
99
+ "joint_limit_active_step_rate": 0.015568520514934365,
100
+ "finger_joint_limit_saturation_step_rate": 0.002367514321347792,
101
+ "max_joint_limit_excess_rad": 0.36301761865615845,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_index_MCP_FE": 662,
105
+ "left_index_PIP": 862,
106
+ "multi_right_index_PIP": 12
107
+ },
108
+ "joint_limit_hard_counts": {
109
+ "left_index_MCP_FE": 22,
110
+ "left_index_PIP": 543,
111
+ "left_pinky_CMC": 2,
112
+ "multi_right_index_PIP": 20
113
+ },
114
+ "joint_limit_hard_episode_events": 519,
115
+ "finger_joint_limit_saturation_counts": {
116
+ "left_index_MCP_FE": 220,
117
+ "left_index_PIP": 4
118
+ },
119
+ "joint_limit_max_excess_by_joint": {
120
+ "A7": 0.0003962516784667969,
121
+ "left_index_DIP": 1.3173590559745207e-05,
122
+ "left_index_MCP_FE": 0.36301761865615845,
123
+ "left_index_PIP": 0.20490120351314545,
124
+ "left_middle_DIP": 0.0001264810562133789,
125
+ "left_middle_MCP_FE": 0.0002536773681640625,
126
+ "left_middle_PIP": 0.001802206039428711,
127
+ "left_pinky_CMC": 0.005450102966278791,
128
+ "left_pinky_DIP": 0.00024628639221191406,
129
+ "left_pinky_MCP_FE": 0.00017893314361572266,
130
+ "left_pinky_PIP": 0.0014020204544067383,
131
+ "left_ring_DIP": 0.00016796588897705078,
132
+ "left_ring_MCP_FE": 8.869171142578125e-05,
133
+ "left_ring_PIP": 0.0012820959091186523,
134
+ "left_thumb_IP": 6.110515565760366e-11,
135
+ "multi_right_index_DIP": 1.2932424397149589e-05,
136
+ "multi_right_index_PIP": 0.07846953719854355,
137
+ "multi_right_middle_DIP": 6.80685043334961e-05,
138
+ "multi_right_middle_MCP_FE": 0.00020623207092285156,
139
+ "multi_right_middle_PIP": 0.0014969110488891602,
140
+ "multi_right_pinky_CMC": 0.0015400053234770894,
141
+ "multi_right_pinky_DIP": 0.0001838207244873047,
142
+ "multi_right_pinky_MCP_FE": 0.0005848407745361328,
143
+ "multi_right_pinky_PIP": 0.00046074390411376953,
144
+ "multi_right_ring_DIP": 0.00010943412780761719,
145
+ "multi_right_ring_MCP_FE": 0.0002199411392211914,
146
+ "multi_right_ring_PIP": 0.0006940364837646484,
147
+ "multi_right_thumb_IP": 1.8656653999471473e-10
148
+ },
149
+ "coverage": {
150
+ "hard_limit_source": {
151
+ "available": 26,
152
+ "unavailable": 0,
153
+ "unknown": 0
154
+ },
155
+ "joint_role_manifest": {
156
+ "available": 26,
157
+ "unavailable": 0,
158
+ "unknown": 0
159
+ }
160
+ },
161
+ "hard_limit_sources": {
162
+ "runtime.joint_pos_limits": 26
163
+ },
164
+ "soft_limit_sources": {
165
+ "runtime.soft_joint_pos_limits": 26
166
+ },
167
+ "unknown_runtime_joint_names": [],
168
+ "baseline_checks": {
169
+ "post_home": {
170
+ "checked": 26,
171
+ "hard_violation_episodes": 0
172
+ },
173
+ "post_reset": {
174
+ "checked": 26,
175
+ "hard_violation_episodes": 0
176
+ }
177
+ }
178
+ }
179
+ },
180
+ "robot_motion": {
181
+ "joint_acceleration_rms": 17.411644829182332,
182
+ "joint_effort_rms": 31.713839258406992,
183
+ "joint_jerk_rms": 1802.4459950063751,
184
+ "joint_velocity_rms": 0.5175245183795028
185
+ },
186
+ "grasp": {
187
+ "kinematic_grasp_stability_index": 0.04193021922383149,
188
+ "mean_kinematic_grasp_stability": 0.30041010325981593,
189
+ "per_object_best_kinematic_grasp_stability": {
190
+ "obj_326_piece_1": 0.0,
191
+ "obj_326_piece_2": 0.002772122129527633,
192
+ "obj_326_piece_3": 0.03837216719223925,
193
+ "obj_326_piece_4": 0.0,
194
+ "obj_326_piece_5": 0.0007859299020646055
195
+ },
196
+ "grasp_gsi_diagnostics": {
197
+ "mean_anti_slip_score": 0.9981531979058759,
198
+ "mean_duration_score": 0.006653349398609085,
199
+ "mean_hold_score": 0.0012321643731371663,
200
+ "mean_kinematic_stability_score": 0.0011476384206304882,
201
+ "mean_legacy_gsi": 0.2016979604890293,
202
+ "mean_lift_score": 0.001388800811719196,
203
+ "mean_motion_stability_score": 0.001214345730523996
204
+ }
205
+ },
206
+ "tool": {
207
+ "tool_selection_accuracy": null,
208
+ "tool_switch_success_rate": null,
209
+ "tool_switch_total": 0,
210
+ "tool_switch_successful": 0
211
+ },
212
+ "runtime": {
213
+ "avg_steps": 3639,
214
+ "error_count": 0
215
+ }
216
+ }
217
+ }
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/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/73_jigsaw_puzzle_assembly/Inv/native_shards/001/summary.json ADDED
@@ -0,0 +1,218 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA yellowbounds-df764f33e4",
5
+ "base_seed": 107300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 0,
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": 3639,
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": 0.0,
50
+ "success_rate_ci": {
51
+ "mean": 0.0,
52
+ "ci_low": 0.0,
53
+ "ci_high": 0.0
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.36
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": null
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.0,
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": 0.0,
74
+ "at_end_success_rate": 0.0,
75
+ "task_family_success_rate": {
76
+ "tetris_block_manipulation": 0.0
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.23,
81
+ "current_chain_depth_progress_score": 0.72,
82
+ "latched_chain_depth_progress_score": 0.88,
83
+ "avg_current_chain_depth": 0.72,
84
+ "avg_latched_chain_depth": 0.88
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": null,
88
+ "expert_time_s": 40.43333333333333,
89
+ "expert_time_step": 2426,
90
+ "avg_steps_to_success": null,
91
+ "avg_policy_steps_to_success": null,
92
+ "avg_policy_queries_to_success": null
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": 0.64,
99
+ "joint_limit_active_step_rate": 0.016257213520197857,
100
+ "finger_joint_limit_saturation_step_rate": 0.003660346248969497,
101
+ "max_joint_limit_excess_rad": 0.3878782391548157,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_index_MCP_FE": 541,
105
+ "left_index_PIP": 982,
106
+ "left_pinky_CMC": 1,
107
+ "multi_right_index_PIP": 23
108
+ },
109
+ "joint_limit_hard_counts": {
110
+ "left_index_MCP_FE": 26,
111
+ "left_index_PIP": 466,
112
+ "left_pinky_CMC": 11,
113
+ "multi_right_index_PIP": 14
114
+ },
115
+ "joint_limit_hard_episode_events": 437,
116
+ "finger_joint_limit_saturation_counts": {
117
+ "left_index_MCP_FE": 263,
118
+ "left_index_PIP": 70
119
+ },
120
+ "joint_limit_max_excess_by_joint": {
121
+ "A7": 0.001008749008178711,
122
+ "left_index_DIP": 1.7503225535620004e-05,
123
+ "left_index_MCP_FE": 0.3878782391548157,
124
+ "left_index_PIP": 0.3290659487247467,
125
+ "left_middle_DIP": 0.00017881393432617188,
126
+ "left_middle_MCP_FE": 0.0002636909484863281,
127
+ "left_middle_PIP": 0.0026388168334960938,
128
+ "left_pinky_CMC": 0.013844229280948639,
129
+ "left_pinky_DIP": 0.00024187564849853516,
130
+ "left_pinky_MCP_FE": 0.00013756752014160156,
131
+ "left_pinky_PIP": 0.0014891624450683594,
132
+ "left_ring_DIP": 0.00013947486877441406,
133
+ "left_ring_MCP_FE": 0.00011992454528808594,
134
+ "left_ring_PIP": 0.0012714862823486328,
135
+ "left_thumb_IP": 1.8638417198513224e-10,
136
+ "multi_right_index_DIP": 6.581045909115346e-06,
137
+ "multi_right_index_PIP": 0.1647748202085495,
138
+ "multi_right_middle_DIP": 6.532669067382812e-05,
139
+ "multi_right_middle_MCP_FE": 0.0002543926239013672,
140
+ "multi_right_middle_PIP": 0.0013788938522338867,
141
+ "multi_right_pinky_CMC": 0.0022097621113061905,
142
+ "multi_right_pinky_DIP": 0.0002741813659667969,
143
+ "multi_right_pinky_MCP_FE": 0.0007528066635131836,
144
+ "multi_right_pinky_PIP": 0.0004832744598388672,
145
+ "multi_right_ring_DIP": 0.0001461505889892578,
146
+ "multi_right_ring_MCP_FE": 0.0002397298812866211,
147
+ "multi_right_ring_PIP": 0.0004411935806274414,
148
+ "multi_right_thumb_IP": 1.0271324901722778e-09
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": 17.368573652052888,
183
+ "joint_effort_rms": 31.68301175347706,
184
+ "joint_jerk_rms": 1796.9604967653077,
185
+ "joint_velocity_rms": 0.5206720204919932
186
+ },
187
+ "grasp": {
188
+ "kinematic_grasp_stability_index": 0.1487275544312639,
189
+ "mean_kinematic_grasp_stability": 0.29690886415435375,
190
+ "per_object_best_kinematic_grasp_stability": {
191
+ "obj_326_piece_1": 0.0,
192
+ "obj_326_piece_2": 0.03900509900307583,
193
+ "obj_326_piece_3": 0.08323695018844421,
194
+ "obj_326_piece_4": 0.0,
195
+ "obj_326_piece_5": 0.026485505239743842
196
+ },
197
+ "grasp_gsi_diagnostics": {
198
+ "mean_anti_slip_score": 0.9985453879270846,
199
+ "mean_duration_score": 0.010742584959238053,
200
+ "mean_hold_score": 0.0012100027480076924,
201
+ "mean_kinematic_stability_score": 0.0010438272343537502,
202
+ "mean_legacy_gsi": 0.20255952361451568,
203
+ "mean_lift_score": 0.0014883209672987084,
204
+ "mean_motion_stability_score": 0.0011207101556679772
205
+ }
206
+ },
207
+ "tool": {
208
+ "tool_selection_accuracy": null,
209
+ "tool_switch_success_rate": null,
210
+ "tool_switch_total": 0,
211
+ "tool_switch_successful": 0
212
+ },
213
+ "runtime": {
214
+ "avg_steps": 3639,
215
+ "error_count": 0
216
+ }
217
+ }
218
+ }
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Inv/per_episode.jsonl ADDED
The diff for this file is too large to render. See raw diff
 
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/Inv/summary.json ADDED
@@ -0,0 +1,218 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metric_schema_version": "core-v1.8-policy-step-query-split",
3
+ "metadata": {
4
+ "policy_name": "PyRUA yellowbounds-df764f33e4",
5
+ "base_seed": 107300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 50,
9
+ "successes": 0,
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": 3639,
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": 0.0,
50
+ "success_rate_ci": {
51
+ "mean": 0.0,
52
+ "ci_low": 0.0,
53
+ "ci_high": 0.0
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.38
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": null
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.0,
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": 0.0,
74
+ "at_end_success_rate": 0.0,
75
+ "task_family_success_rate": {
76
+ "tetris_block_manipulation": 0.0
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.255,
81
+ "current_chain_depth_progress_score": 0.74,
82
+ "latched_chain_depth_progress_score": 0.92,
83
+ "avg_current_chain_depth": 0.74,
84
+ "avg_latched_chain_depth": 0.92
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": null,
88
+ "expert_time_s": 40.43333333333333,
89
+ "expert_time_step": 2426,
90
+ "avg_steps_to_success": null,
91
+ "avg_policy_steps_to_success": null,
92
+ "avg_policy_queries_to_success": null
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": 0.58,
99
+ "joint_limit_active_step_rate": 0.016180269304754052,
100
+ "finger_joint_limit_saturation_step_rate": 0.0030612805715856005,
101
+ "max_joint_limit_excess_rad": 0.3878782391548157,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_index_MCP_FE": 1196,
105
+ "left_index_PIP": 1839,
106
+ "left_pinky_CMC": 1,
107
+ "multi_right_index_PIP": 35
108
+ },
109
+ "joint_limit_hard_counts": {
110
+ "left_index_MCP_FE": 47,
111
+ "left_index_PIP": 993,
112
+ "left_pinky_CMC": 12,
113
+ "multi_right_index_PIP": 34
114
+ },
115
+ "joint_limit_hard_episode_events": 941,
116
+ "finger_joint_limit_saturation_counts": {
117
+ "left_index_MCP_FE": 483,
118
+ "left_index_PIP": 74
119
+ },
120
+ "joint_limit_max_excess_by_joint": {
121
+ "A7": 0.001008749008178711,
122
+ "left_index_DIP": 1.7503225535620004e-05,
123
+ "left_index_MCP_FE": 0.3878782391548157,
124
+ "left_index_PIP": 0.3290659487247467,
125
+ "left_middle_DIP": 0.00017881393432617188,
126
+ "left_middle_MCP_FE": 0.0002636909484863281,
127
+ "left_middle_PIP": 0.0026388168334960938,
128
+ "left_pinky_CMC": 0.013844229280948639,
129
+ "left_pinky_DIP": 0.00024628639221191406,
130
+ "left_pinky_MCP_FE": 0.00017893314361572266,
131
+ "left_pinky_PIP": 0.0014891624450683594,
132
+ "left_ring_DIP": 0.00016796588897705078,
133
+ "left_ring_MCP_FE": 0.00011992454528808594,
134
+ "left_ring_PIP": 0.0012820959091186523,
135
+ "left_thumb_IP": 1.8638417198513224e-10,
136
+ "multi_right_index_DIP": 1.2932424397149589e-05,
137
+ "multi_right_index_PIP": 0.1647748202085495,
138
+ "multi_right_middle_DIP": 6.80685043334961e-05,
139
+ "multi_right_middle_MCP_FE": 0.0002543926239013672,
140
+ "multi_right_middle_PIP": 0.0014969110488891602,
141
+ "multi_right_pinky_CMC": 0.0022097621113061905,
142
+ "multi_right_pinky_DIP": 0.0002741813659667969,
143
+ "multi_right_pinky_MCP_FE": 0.0007528066635131836,
144
+ "multi_right_pinky_PIP": 0.0004832744598388672,
145
+ "multi_right_ring_DIP": 0.0001461505889892578,
146
+ "multi_right_ring_MCP_FE": 0.0002397298812866211,
147
+ "multi_right_ring_PIP": 0.0006940364837646484,
148
+ "multi_right_thumb_IP": 1.0271324901722778e-09
149
+ },
150
+ "coverage": {
151
+ "hard_limit_source": {
152
+ "available": 50,
153
+ "unavailable": 0,
154
+ "unknown": 0
155
+ },
156
+ "joint_role_manifest": {
157
+ "available": 50,
158
+ "unavailable": 0,
159
+ "unknown": 0
160
+ }
161
+ },
162
+ "hard_limit_sources": {
163
+ "runtime.joint_pos_limits": 50
164
+ },
165
+ "soft_limit_sources": {
166
+ "runtime.soft_joint_pos_limits": 50
167
+ },
168
+ "unknown_runtime_joint_names": [],
169
+ "baseline_checks": {
170
+ "post_home": {
171
+ "checked": 50,
172
+ "hard_violation_episodes": 0
173
+ },
174
+ "post_reset": {
175
+ "checked": 50,
176
+ "hard_violation_episodes": 0
177
+ }
178
+ }
179
+ }
180
+ },
181
+ "robot_motion": {
182
+ "joint_acceleration_rms": 17.405117604101196,
183
+ "joint_effort_rms": 31.722971928248562,
184
+ "joint_jerk_rms": 1801.3802645057788,
185
+ "joint_velocity_rms": 0.5191662804629062
186
+ },
187
+ "grasp": {
188
+ "kinematic_grasp_stability_index": 0.09472598770466994,
189
+ "mean_kinematic_grasp_stability": 0.33151832892718736,
190
+ "per_object_best_kinematic_grasp_stability": {
191
+ "obj_326_piece_1": 0.0,
192
+ "obj_326_piece_2": 0.019502549501537916,
193
+ "obj_326_piece_3": 0.06157200203418652,
194
+ "obj_326_piece_4": 0.0,
195
+ "obj_326_piece_5": 0.013651436168945515
196
+ },
197
+ "grasp_gsi_diagnostics": {
198
+ "mean_anti_slip_score": 0.9983123568745977,
199
+ "mean_duration_score": 0.00883103416689575,
200
+ "mean_hold_score": 0.0012432536411101933,
201
+ "mean_kinematic_stability_score": 0.0011181821026930543,
202
+ "mean_legacy_gsi": 0.20215986404268352,
203
+ "mean_lift_score": 0.0014498488595768068,
204
+ "mean_motion_stability_score": 0.0011884582362953018
205
+ }
206
+ },
207
+ "tool": {
208
+ "tool_selection_accuracy": null,
209
+ "tool_switch_success_rate": null,
210
+ "tool_switch_total": 0,
211
+ "tool_switch_successful": 0
212
+ },
213
+ "runtime": {
214
+ "avg_steps": 3639,
215
+ "error_count": 0
216
+ }
217
+ }
218
+ }
pyrua-dex/per_task_eval/73_jigsaw_puzzle_assembly/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/73_jigsaw_puzzle_assembly/None/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 yellowbounds-df764f33e4",
5
+ "base_seed": 107300000,
6
+ "seed_policy": "episode_seed = base_seed + episode_index - 1",
7
+ "randomization_recorded": true,
8
+ "n_episodes": 25,
9
+ "successes": 0,
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": 3639,
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": 0.0,
50
+ "success_rate_ci": {
51
+ "mean": 0.0,
52
+ "ci_low": 0.0,
53
+ "ci_high": 0.0
54
+ }
55
+ },
56
+ "progress": {
57
+ "latched_stage_completion_rate": 0.38
58
+ },
59
+ "efficiency": {
60
+ "avg_time_to_success_s": null
61
+ },
62
+ "safety": {
63
+ "safe_success_rate": 0.0,
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": 0.0,
74
+ "at_end_success_rate": 0.0,
75
+ "task_family_success_rate": {
76
+ "tetris_block_manipulation": 0.0
77
+ }
78
+ },
79
+ "progress": {
80
+ "current_stage_completion_rate": 0.26,
81
+ "current_chain_depth_progress_score": 0.72,
82
+ "latched_chain_depth_progress_score": 0.88,
83
+ "avg_current_chain_depth": 0.72,
84
+ "avg_latched_chain_depth": 0.88
85
+ },
86
+ "efficiency": {
87
+ "task_efficiency": null,
88
+ "expert_time_s": 40.43333333333333,
89
+ "expert_time_step": 2426,
90
+ "avg_steps_to_success": null,
91
+ "avg_policy_steps_to_success": null,
92
+ "avg_policy_queries_to_success": null
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": 0.56,
99
+ "joint_limit_active_step_rate": 0.012772739763671338,
100
+ "finger_joint_limit_saturation_step_rate": 0.002660071448200055,
101
+ "max_joint_limit_excess_rad": 0.5290722846984863,
102
+ "robot_constraint_diagnostics": {
103
+ "joint_limit_soft_counts": {
104
+ "left_index_DIP": 4,
105
+ "left_index_MCP_FE": 498,
106
+ "left_index_PIP": 705,
107
+ "multi_right_index_DIP": 15,
108
+ "multi_right_index_PIP": 8
109
+ },
110
+ "joint_limit_hard_counts": {
111
+ "left_index_DIP": 3,
112
+ "left_index_MCP_FE": 20,
113
+ "left_index_PIP": 525,
114
+ "left_pinky_CMC": 3,
115
+ "multi_right_index_DIP": 1,
116
+ "multi_right_index_PIP": 13
117
+ },
118
+ "joint_limit_hard_episode_events": 503,
119
+ "finger_joint_limit_saturation_counts": {
120
+ "left_index_MCP_FE": 217,
121
+ "left_index_PIP": 17,
122
+ "multi_right_index_DIP": 8
123
+ },
124
+ "joint_limit_max_excess_by_joint": {
125
+ "A7": 0.001058816909790039,
126
+ "left_index_DIP": 0.023953793570399284,
127
+ "left_index_MCP_FE": 0.3745415210723877,
128
+ "left_index_PIP": 0.5290722846984863,
129
+ "left_middle_DIP": 0.00011396408081054688,
130
+ "left_middle_MCP_FE": 0.0002256631851196289,
131
+ "left_middle_PIP": 0.0026121139526367188,
132
+ "left_pinky_CMC": 0.005511615891009569,
133
+ "left_pinky_DIP": 0.0002751350402832031,
134
+ "left_pinky_MCP_FE": 0.00012409687042236328,
135
+ "left_pinky_PIP": 0.0015393495559692383,
136
+ "left_ring_DIP": 0.00013506412506103516,
137
+ "left_ring_MCP_FE": 7.724761962890625e-05,
138
+ "left_ring_PIP": 0.0013428926467895508,
139
+ "left_thumb_IP": 8.207264728943287e-11,
140
+ "multi_right_index_DIP": 0.45763352513313293,
141
+ "multi_right_index_PIP": 0.029717331752181053,
142
+ "multi_right_middle_DIP": 2.300739288330078e-05,
143
+ "multi_right_middle_MCP_FE": 0.00014841556549072266,
144
+ "multi_right_middle_PIP": 0.0014123916625976562,
145
+ "multi_right_pinky_CMC": 0.0028947184327989817,
146
+ "multi_right_pinky_DIP": 0.00012254714965820312,
147
+ "multi_right_pinky_MCP_FE": 0.000574946403503418,
148
+ "multi_right_pinky_PIP": 0.0011763572692871094,
149
+ "multi_right_ring_DIP": 5.650520324707031e-05,
150
+ "multi_right_ring_MCP_FE": 0.00019502639770507812,
151
+ "multi_right_ring_PIP": 0.001379251480102539,
152
+ "multi_right_thumb_IP": 1.7997987822315764e-10
153
+ },
154
+ "coverage": {
155
+ "hard_limit_source": {
156
+ "available": 25,
157
+ "unavailable": 0,
158
+ "unknown": 0
159
+ },
160
+ "joint_role_manifest": {
161
+ "available": 25,
162
+ "unavailable": 0,
163
+ "unknown": 0
164
+ }
165
+ },
166
+ "hard_limit_sources": {
167
+ "runtime.joint_pos_limits": 25
168
+ },
169
+ "soft_limit_sources": {
170
+ "runtime.soft_joint_pos_limits": 25
171
+ },
172
+ "unknown_runtime_joint_names": [],
173
+ "baseline_checks": {
174
+ "post_home": {
175
+ "checked": 25,
176
+ "hard_violation_episodes": 0
177
+ },
178
+ "post_reset": {
179
+ "checked": 25,
180
+ "hard_violation_episodes": 0
181
+ }
182
+ }
183
+ }
184
+ },
185
+ "robot_motion": {
186
+ "joint_acceleration_rms": 17.391501217130923,
187
+ "joint_effort_rms": 31.981414458437655,
188
+ "joint_jerk_rms": 1800.5410420784,
189
+ "joint_velocity_rms": 0.521150143535972
190
+ },
191
+ "grasp": {
192
+ "kinematic_grasp_stability_index": 0.07871903262382561,
193
+ "mean_kinematic_grasp_stability": 0.09021379820219946,
194
+ "per_object_best_kinematic_grasp_stability": {
195
+ "obj_326_piece_1": 0.0,
196
+ "obj_326_piece_2": 0.0,
197
+ "obj_326_piece_3": 0.05962116954113134,
198
+ "obj_326_piece_4": 0.0,
199
+ "obj_326_piece_5": 0.019097863082694275
200
+ },
201
+ "grasp_gsi_diagnostics": {
202
+ "mean_anti_slip_score": 0.9977004671613067,
203
+ "mean_duration_score": 0.0035812769075753374,
204
+ "mean_hold_score": 9.20032976092332e-05,
205
+ "mean_kinematic_stability_score": 3.518094776428069e-05,
206
+ "mean_legacy_gsi": 0.20033733757078964,
207
+ "mean_lift_score": 0.0003781258587524045,
208
+ "mean_motion_stability_score": 0.0001028518897905106
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": 3639,
219
+ "error_count": 0
220
+ }
221
+ }
222
+ }