auto-sync 2026-07-03T17:25:27Z workspace (part 9)
Browse files- workspace/runs/ctt_test_rollout_comparison/combined_measured_candidates.json +2 -0
- workspace/runs/ctt_test_rollout_comparison/command.txt +1 -1
- workspace/runs/ctt_test_rollout_comparison/measured_metrics/metrics.json +0 -0
- workspace/runs/ctt_test_rollout_comparison/measured_metrics/metrics_by_seed.json +0 -0
- workspace/runs/ctt_test_rollout_comparison/measured_metrics/metrics_by_task.json +20 -0
- workspace/runs/ctt_test_rollout_comparison/measured_metrics/report.md +21 -17
- workspace/runs/ctt_test_rollout_comparison/measured_metrics/table.tex +21 -17
- workspace/runs/ctt_test_rollout_comparison/metrics.json +856 -195
- workspace/runs/ctt_test_rollout_comparison/metrics_by_seed.json +0 -0
- workspace/runs/ctt_test_rollout_comparison/metrics_by_task.json +20 -0
- workspace/runs/ctt_test_rollout_comparison/report.md +30 -17
- workspace/runs/ctt_test_rollout_comparison/table.tex +21 -17
- workspace/runs/ctt_val_proxy_comparison/eval.log +2 -0
- workspace/runs/ctt_val_proxy_comparison/metrics.json +36 -0
- workspace/runs/ctt_val_proxy_comparison/report.md +2 -0
- workspace/runs/ctt_val_proxy_comparison/table.tex +2 -0
- workspace/runs/ctt_val_rollout_comparison/combined_measured_candidates.json +2 -0
- workspace/runs/ctt_val_rollout_comparison/command.txt +1 -1
- workspace/runs/ctt_val_rollout_comparison/measured_metrics/metrics.json +151 -151
- workspace/runs/ctt_val_rollout_comparison/measured_metrics/report.md +18 -18
- workspace/runs/ctt_val_rollout_comparison/measured_metrics/table.tex +18 -18
- workspace/runs/ctt_val_rollout_comparison/metrics.json +160 -151
- workspace/runs/ctt_val_rollout_comparison/report.md +27 -18
- workspace/runs/ctt_val_rollout_comparison/table.tex +18 -18
- workspace/runs/leakage_audit_rgb_refs_object/index_report.json +35 -0
- workspace/runs/leakage_audit_rgb_refs_object/report.json +159 -0
- workspace/runs/leakage_audit_rgb_refs_object/report.md +136 -0
- workspace/runs/summary_ctt.csv +22 -0
- workspace/runs/summary_ctt.md +22 -0
workspace/runs/ctt_test_rollout_comparison/combined_measured_candidates.json
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
{
|
|
|
|
| 2 |
"k": 8,
|
| 3 |
"num_rows": 144,
|
| 4 |
"report_type": "ctt_measured_rollout_comparison",
|
|
@@ -97217,6 +97218,7 @@
|
|
| 97217 |
],
|
| 97218 |
"schema_version": 1,
|
| 97219 |
"source_content_hash": "90b431047a1f35cec095cbafa7de5bc0126cb0d0f4db3ff31843dca369977957",
|
|
|
|
| 97220 |
"target_content_hash": "1710ede7c73d8f72479fd63dd9941a0e8bb55211d3ea0267963abc9e7a043b91",
|
| 97221 |
"target_split_hash": "84151df9e88c219b76751677558cc5bbbc19082bf358cea7330158e9e403e25f",
|
| 97222 |
"train_seeds": [
|
|
|
|
| 1 |
{
|
| 2 |
+
"data_hash": "1710ede7c73d8f72479fd63dd9941a0e8bb55211d3ea0267963abc9e7a043b91",
|
| 3 |
"k": 8,
|
| 4 |
"num_rows": 144,
|
| 5 |
"report_type": "ctt_measured_rollout_comparison",
|
|
|
|
| 97218 |
],
|
| 97219 |
"schema_version": 1,
|
| 97220 |
"source_content_hash": "90b431047a1f35cec095cbafa7de5bc0126cb0d0f4db3ff31843dca369977957",
|
| 97221 |
+
"split_hash": "84151df9e88c219b76751677558cc5bbbc19082bf358cea7330158e9e403e25f",
|
| 97222 |
"target_content_hash": "1710ede7c73d8f72479fd63dd9941a0e8bb55211d3ea0267963abc9e7a043b91",
|
| 97223 |
"target_split_hash": "84151df9e88c219b76751677558cc5bbbc19082bf358cea7330158e9e403e25f",
|
| 97224 |
"train_seeds": [
|
workspace/runs/ctt_test_rollout_comparison/command.txt
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
python scripts/build_ctt_rollout_comparison.py --run-glob runs/ctt_residual_rollout_test_seed* --out-dir runs/ctt_test_rollout_comparison --k 8
|
|
|
|
| 1 |
+
python scripts/build_ctt_rollout_comparison.py --run-glob runs/ctt_residual_rollout_test_seed* --out-dir runs/ctt_test_rollout_comparison --k 8
|
workspace/runs/ctt_test_rollout_comparison/measured_metrics/metrics.json
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
workspace/runs/ctt_test_rollout_comparison/measured_metrics/metrics_by_seed.json
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
workspace/runs/ctt_test_rollout_comparison/measured_metrics/metrics_by_task.json
CHANGED
|
@@ -1,16 +1,20 @@
|
|
| 1 |
{
|
| 2 |
"LiftPegUpright-v1": {
|
|
|
|
| 3 |
"base_success": 0.2,
|
| 4 |
"base_utility": 0.9498664379119873,
|
| 5 |
"branch_car_at_8": 0.5633442640304566,
|
|
|
|
| 6 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 7 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 8 |
"outcome_ptr_at_8": 0.7333333333333333,
|
| 9 |
"pairwise_causal_calibration_ece": 0.1621114782545992,
|
|
|
|
| 10 |
"proposal_oracle_success_at_8": 0.6666666666666666,
|
| 11 |
"proposal_oracle_success_gain_over_base_at_8": 0.4666666666666667,
|
| 12 |
"proposal_oracle_utility_at_8": 1.5397698879241943,
|
| 13 |
"proposal_oracle_utility_gain_over_base_at_8": 0.589903450012207,
|
|
|
|
| 14 |
"selected_success_at_8": 0.3333333333333333,
|
| 15 |
"selected_success_gain_over_base_at_8": 0.13333333333333333,
|
| 16 |
"selected_utility_at_8": 0.9764256238937378,
|
|
@@ -23,17 +27,21 @@
|
|
| 23 |
"total_car_to_hidden_at_8": 1.0235743761062621
|
| 24 |
},
|
| 25 |
"PickCube-v1": {
|
|
|
|
| 26 |
"base_success": 0.0,
|
| 27 |
"base_utility": 0.5839591026306152,
|
| 28 |
"branch_car_at_8": 0.15916230529546738,
|
|
|
|
| 29 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 30 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 31 |
"outcome_ptr_at_8": 0.0,
|
| 32 |
"pairwise_causal_calibration_ece": 0.2290622239534128,
|
|
|
|
| 33 |
"proposal_oracle_success_at_8": 0.0,
|
| 34 |
"proposal_oracle_success_gain_over_base_at_8": 0.0,
|
| 35 |
"proposal_oracle_utility_at_8": 0.25069022675355274,
|
| 36 |
"proposal_oracle_utility_gain_over_base_at_8": -0.3332688758770625,
|
|
|
|
| 37 |
"selected_success_at_8": 0.0,
|
| 38 |
"selected_success_gain_over_base_at_8": 0.0,
|
| 39 |
"selected_utility_at_8": 0.09152792145808537,
|
|
@@ -46,17 +54,21 @@
|
|
| 46 |
"total_car_to_hidden_at_8": 1.9084720785419147
|
| 47 |
},
|
| 48 |
"PullCube-v1": {
|
|
|
|
| 49 |
"base_success": 0.37333333333333335,
|
| 50 |
"base_utility": 0.7993177033324416,
|
| 51 |
"branch_car_at_8": 0.40464685219650465,
|
|
|
|
| 52 |
"hidden_chart_oracle_success_at_8": 0.52,
|
| 53 |
"hidden_chart_oracle_utility_at_8": 1.1467878860235214,
|
| 54 |
"outcome_ptr_at_8": 0.38666666666666666,
|
| 55 |
"pairwise_causal_calibration_ece": 0.17280987267234024,
|
|
|
|
| 56 |
"proposal_oracle_success_at_8": 0.4533333333333333,
|
| 57 |
"proposal_oracle_success_gain_over_base_at_8": 0.08,
|
| 58 |
"proposal_oracle_utility_at_8": 0.9498632888682187,
|
| 59 |
"proposal_oracle_utility_gain_over_base_at_8": 0.1505455855357771,
|
|
|
|
| 60 |
"selected_success_at_8": 0.25333333333333335,
|
| 61 |
"selected_success_gain_over_base_at_8": -0.12,
|
| 62 |
"selected_utility_at_8": 0.5452164366717139,
|
|
@@ -69,17 +81,21 @@
|
|
| 69 |
"total_car_to_hidden_at_8": 0.6015714493518074
|
| 70 |
},
|
| 71 |
"PushCube-v1": {
|
|
|
|
| 72 |
"base_success": 0.2222222222222222,
|
| 73 |
"base_utility": 0.49637971735662884,
|
| 74 |
"branch_car_at_8": 1.178912353184488,
|
|
|
|
| 75 |
"hidden_chart_oracle_success_at_8": 0.8333333333333334,
|
| 76 |
"hidden_chart_oracle_utility_at_8": 1.7015800600250561,
|
| 77 |
"outcome_ptr_at_8": 0.7777777777777778,
|
| 78 |
"pairwise_causal_calibration_ece": 0.1716092477675637,
|
|
|
|
| 79 |
"proposal_oracle_success_at_8": 0.8333333333333334,
|
| 80 |
"proposal_oracle_success_gain_over_base_at_8": 0.6111111111111112,
|
| 81 |
"proposal_oracle_utility_at_8": 1.6836495283577178,
|
| 82 |
"proposal_oracle_utility_gain_over_base_at_8": 1.1872698110010889,
|
|
|
|
| 83 |
"selected_success_at_8": 0.2222222222222222,
|
| 84 |
"selected_success_gain_over_base_at_8": 0.0,
|
| 85 |
"selected_utility_at_8": 0.5047371751732297,
|
|
@@ -92,17 +108,21 @@
|
|
| 92 |
"total_car_to_hidden_at_8": 1.1968428848518267
|
| 93 |
},
|
| 94 |
"StackCube-v1": {
|
|
|
|
| 95 |
"base_success": 0.18181818181818182,
|
| 96 |
"base_utility": 0.6905394485502532,
|
| 97 |
"branch_car_at_8": 0.6124884856469703,
|
|
|
|
| 98 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 99 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 100 |
"outcome_ptr_at_8": 0.6666666666666666,
|
| 101 |
"pairwise_causal_calibration_ece": 0.1741911827659335,
|
|
|
|
| 102 |
"proposal_oracle_success_at_8": 0.42424242424242425,
|
| 103 |
"proposal_oracle_success_gain_over_base_at_8": 0.24242424242424243,
|
| 104 |
"proposal_oracle_utility_at_8": 1.0394560405702302,
|
| 105 |
"proposal_oracle_utility_gain_over_base_at_8": 0.348916592019977,
|
|
|
|
| 106 |
"selected_success_at_8": 0.12121212121212122,
|
| 107 |
"selected_success_gain_over_base_at_8": -0.06060606060606061,
|
| 108 |
"selected_utility_at_8": 0.42696755492325983,
|
|
|
|
| 1 |
{
|
| 2 |
"LiftPegUpright-v1": {
|
| 3 |
+
"base_safety_label_known": 0.0,
|
| 4 |
"base_success": 0.2,
|
| 5 |
"base_utility": 0.9498664379119873,
|
| 6 |
"branch_car_at_8": 0.5633442640304566,
|
| 7 |
+
"generated_safety_label_coverage_at_8": 0.0,
|
| 8 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 9 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 10 |
"outcome_ptr_at_8": 0.7333333333333333,
|
| 11 |
"pairwise_causal_calibration_ece": 0.1621114782545992,
|
| 12 |
+
"proposal_oracle_safety_label_known_at_8": 0.0,
|
| 13 |
"proposal_oracle_success_at_8": 0.6666666666666666,
|
| 14 |
"proposal_oracle_success_gain_over_base_at_8": 0.4666666666666667,
|
| 15 |
"proposal_oracle_utility_at_8": 1.5397698879241943,
|
| 16 |
"proposal_oracle_utility_gain_over_base_at_8": 0.589903450012207,
|
| 17 |
+
"selected_safety_label_known_at_8": 0.0,
|
| 18 |
"selected_success_at_8": 0.3333333333333333,
|
| 19 |
"selected_success_gain_over_base_at_8": 0.13333333333333333,
|
| 20 |
"selected_utility_at_8": 0.9764256238937378,
|
|
|
|
| 27 |
"total_car_to_hidden_at_8": 1.0235743761062621
|
| 28 |
},
|
| 29 |
"PickCube-v1": {
|
| 30 |
+
"base_safety_label_known": 0.0,
|
| 31 |
"base_success": 0.0,
|
| 32 |
"base_utility": 0.5839591026306152,
|
| 33 |
"branch_car_at_8": 0.15916230529546738,
|
| 34 |
+
"generated_safety_label_coverage_at_8": 0.0,
|
| 35 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 36 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 37 |
"outcome_ptr_at_8": 0.0,
|
| 38 |
"pairwise_causal_calibration_ece": 0.2290622239534128,
|
| 39 |
+
"proposal_oracle_safety_label_known_at_8": 0.0,
|
| 40 |
"proposal_oracle_success_at_8": 0.0,
|
| 41 |
"proposal_oracle_success_gain_over_base_at_8": 0.0,
|
| 42 |
"proposal_oracle_utility_at_8": 0.25069022675355274,
|
| 43 |
"proposal_oracle_utility_gain_over_base_at_8": -0.3332688758770625,
|
| 44 |
+
"selected_safety_label_known_at_8": 0.0,
|
| 45 |
"selected_success_at_8": 0.0,
|
| 46 |
"selected_success_gain_over_base_at_8": 0.0,
|
| 47 |
"selected_utility_at_8": 0.09152792145808537,
|
|
|
|
| 54 |
"total_car_to_hidden_at_8": 1.9084720785419147
|
| 55 |
},
|
| 56 |
"PullCube-v1": {
|
| 57 |
+
"base_safety_label_known": 0.0,
|
| 58 |
"base_success": 0.37333333333333335,
|
| 59 |
"base_utility": 0.7993177033324416,
|
| 60 |
"branch_car_at_8": 0.40464685219650465,
|
| 61 |
+
"generated_safety_label_coverage_at_8": 0.0,
|
| 62 |
"hidden_chart_oracle_success_at_8": 0.52,
|
| 63 |
"hidden_chart_oracle_utility_at_8": 1.1467878860235214,
|
| 64 |
"outcome_ptr_at_8": 0.38666666666666666,
|
| 65 |
"pairwise_causal_calibration_ece": 0.17280987267234024,
|
| 66 |
+
"proposal_oracle_safety_label_known_at_8": 0.0,
|
| 67 |
"proposal_oracle_success_at_8": 0.4533333333333333,
|
| 68 |
"proposal_oracle_success_gain_over_base_at_8": 0.08,
|
| 69 |
"proposal_oracle_utility_at_8": 0.9498632888682187,
|
| 70 |
"proposal_oracle_utility_gain_over_base_at_8": 0.1505455855357771,
|
| 71 |
+
"selected_safety_label_known_at_8": 0.0,
|
| 72 |
"selected_success_at_8": 0.25333333333333335,
|
| 73 |
"selected_success_gain_over_base_at_8": -0.12,
|
| 74 |
"selected_utility_at_8": 0.5452164366717139,
|
|
|
|
| 81 |
"total_car_to_hidden_at_8": 0.6015714493518074
|
| 82 |
},
|
| 83 |
"PushCube-v1": {
|
| 84 |
+
"base_safety_label_known": 0.0,
|
| 85 |
"base_success": 0.2222222222222222,
|
| 86 |
"base_utility": 0.49637971735662884,
|
| 87 |
"branch_car_at_8": 1.178912353184488,
|
| 88 |
+
"generated_safety_label_coverage_at_8": 0.0,
|
| 89 |
"hidden_chart_oracle_success_at_8": 0.8333333333333334,
|
| 90 |
"hidden_chart_oracle_utility_at_8": 1.7015800600250561,
|
| 91 |
"outcome_ptr_at_8": 0.7777777777777778,
|
| 92 |
"pairwise_causal_calibration_ece": 0.1716092477675637,
|
| 93 |
+
"proposal_oracle_safety_label_known_at_8": 0.0,
|
| 94 |
"proposal_oracle_success_at_8": 0.8333333333333334,
|
| 95 |
"proposal_oracle_success_gain_over_base_at_8": 0.6111111111111112,
|
| 96 |
"proposal_oracle_utility_at_8": 1.6836495283577178,
|
| 97 |
"proposal_oracle_utility_gain_over_base_at_8": 1.1872698110010889,
|
| 98 |
+
"selected_safety_label_known_at_8": 0.0,
|
| 99 |
"selected_success_at_8": 0.2222222222222222,
|
| 100 |
"selected_success_gain_over_base_at_8": 0.0,
|
| 101 |
"selected_utility_at_8": 0.5047371751732297,
|
|
|
|
| 108 |
"total_car_to_hidden_at_8": 1.1968428848518267
|
| 109 |
},
|
| 110 |
"StackCube-v1": {
|
| 111 |
+
"base_safety_label_known": 0.0,
|
| 112 |
"base_success": 0.18181818181818182,
|
| 113 |
"base_utility": 0.6905394485502532,
|
| 114 |
"branch_car_at_8": 0.6124884856469703,
|
| 115 |
+
"generated_safety_label_coverage_at_8": 0.0,
|
| 116 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 117 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 118 |
"outcome_ptr_at_8": 0.6666666666666666,
|
| 119 |
"pairwise_causal_calibration_ece": 0.1741911827659335,
|
| 120 |
+
"proposal_oracle_safety_label_known_at_8": 0.0,
|
| 121 |
"proposal_oracle_success_at_8": 0.42424242424242425,
|
| 122 |
"proposal_oracle_success_gain_over_base_at_8": 0.24242424242424243,
|
| 123 |
"proposal_oracle_utility_at_8": 1.0394560405702302,
|
| 124 |
"proposal_oracle_utility_gain_over_base_at_8": 0.348916592019977,
|
| 125 |
+
"selected_safety_label_known_at_8": 0.0,
|
| 126 |
"selected_success_at_8": 0.12121212121212122,
|
| 127 |
"selected_success_gain_over_base_at_8": -0.06060606060606061,
|
| 128 |
"selected_utility_at_8": 0.42696755492325983,
|
workspace/runs/ctt_test_rollout_comparison/measured_metrics/report.md
CHANGED
|
@@ -4,24 +4,28 @@ K: `8`
|
|
| 4 |
|
| 5 |
| Metric | N | Micro mean | 95% CI | Task macro | Seed macro |
|
| 6 |
| --- | ---: | ---: | ---: | ---: | ---: |
|
|
|
|
| 7 |
| base_success | 144 | 0.2847 | [0.2153, 0.3542] | 0.1955 | 0.2917 |
|
| 8 |
-
| base_utility | 144 | 0.7477 | [0.
|
| 9 |
-
| branch_car_at_8 | 144 | 0.5605 | [0.
|
| 10 |
-
|
|
| 11 |
-
|
|
| 12 |
-
|
|
| 13 |
-
|
|
| 14 |
-
|
|
| 15 |
-
|
|
| 16 |
-
|
|
| 17 |
-
|
|
|
|
|
|
|
|
|
|
|
| 18 |
| selected_success_at_8 | 144 | 0.2222 | [0.1528, 0.2917] | 0.1860 | 0.2306 |
|
| 19 |
-
| selected_success_gain_over_base_at_8 | 144 | -0.0625 | [-0.
|
| 20 |
-
| selected_utility_at_8 | 144 | 0.5485 | [0.
|
| 21 |
-
| selected_utility_gain_over_base_at_8 | 144 | -0.1992 | [-0.3477, -0.
|
| 22 |
-
| selector_regret_at_8 | 144 | 0.5605 | [0.
|
| 23 |
-
| success_selector_gap_at_8 | 144 | 0.2847 | [0.
|
| 24 |
| success_support_gap_at_8 | 144 | 0.2639 | [0.1944, 0.3333] | 0.4365 | 0.2917 |
|
| 25 |
| success_total_car_to_hidden_at_8 | 144 | 0.5069 | [0.4236, 0.5903] | 0.6846 | 0.5549 |
|
| 26 |
-
| support_gap_at_8 | 144 | 0.4833 | [0.
|
| 27 |
-
| total_car_to_hidden_at_8 | 144 | 0.9698 | [0.
|
|
|
|
| 4 |
|
| 5 |
| Metric | N | Micro mean | 95% CI | Task macro | Seed macro |
|
| 6 |
| --- | ---: | ---: | ---: | ---: | ---: |
|
| 7 |
+
| base_safety_label_known | 144 | 0.0000 | [0.0000, 0.0000] | 0.0000 | 0.0000 |
|
| 8 |
| base_success | 144 | 0.2847 | [0.2153, 0.3542] | 0.1955 | 0.2917 |
|
| 9 |
+
| base_utility | 144 | 0.7477 | [0.6173, 0.8788] | 0.7040 | 0.7857 |
|
| 10 |
+
| branch_car_at_8 | 144 | 0.5605 | [0.4345, 0.7065] | 0.5837 | 0.5889 |
|
| 11 |
+
| generated_safety_label_coverage_at_8 | 144 | 0.0000 | [0.0000, 0.0000] | 0.0000 | 0.0000 |
|
| 12 |
+
| hidden_chart_oracle_success_at_8 | 144 | 0.7292 | [0.6597, 0.7986] | 0.8707 | 0.7854 |
|
| 13 |
+
| hidden_chart_oracle_utility_at_8 | 144 | 1.5183 | [1.3941, 1.6438] | 1.7697 | 1.6191 |
|
| 14 |
+
| outcome_ptr_at_8 | 144 | 0.5278 | [0.4514, 0.6042] | 0.5129 | 0.5243 |
|
| 15 |
+
| pairwise_causal_calibration_ece | 144 | 0.1730 | [0.1547, 0.1911] | 0.1820 | 0.1718 |
|
| 16 |
+
| proposal_oracle_safety_label_known_at_8 | 144 | 0.0000 | [0.0000, 0.0000] | 0.0000 | 0.0000 |
|
| 17 |
+
| proposal_oracle_success_at_8 | 144 | 0.5069 | [0.4306, 0.5903] | 0.4755 | 0.5312 |
|
| 18 |
+
| proposal_oracle_success_gain_over_base_at_8 | 144 | 0.2222 | [0.1319, 0.3056] | 0.2800 | 0.2396 |
|
| 19 |
+
| proposal_oracle_utility_at_8 | 144 | 1.1090 | [0.9685, 1.2662] | 1.0927 | 1.1655 |
|
| 20 |
+
| proposal_oracle_utility_gain_over_base_at_8 | 144 | 0.3613 | [0.2109, 0.5091] | 0.3887 | 0.3798 |
|
| 21 |
+
| selected_safety_label_known_at_8 | 144 | 0.0000 | [0.0000, 0.0000] | 0.0000 | 0.0000 |
|
| 22 |
| selected_success_at_8 | 144 | 0.2222 | [0.1528, 0.2917] | 0.1860 | 0.2306 |
|
| 23 |
+
| selected_success_gain_over_base_at_8 | 144 | -0.0625 | [-0.1458, 0.0139] | -0.0095 | -0.0611 |
|
| 24 |
+
| selected_utility_at_8 | 144 | 0.5485 | [0.4193, 0.6933] | 0.5090 | 0.5766 |
|
| 25 |
+
| selected_utility_gain_over_base_at_8 | 144 | -0.1992 | [-0.3477, -0.0505] | -0.1950 | -0.2092 |
|
| 26 |
+
| selector_regret_at_8 | 144 | 0.5605 | [0.4345, 0.7065] | 0.5837 | 0.5889 |
|
| 27 |
+
| success_selector_gap_at_8 | 144 | 0.2847 | [0.2153, 0.3611] | 0.2895 | 0.3007 |
|
| 28 |
| success_support_gap_at_8 | 144 | 0.2639 | [0.1944, 0.3333] | 0.4365 | 0.2917 |
|
| 29 |
| success_total_car_to_hidden_at_8 | 144 | 0.5069 | [0.4236, 0.5903] | 0.6846 | 0.5549 |
|
| 30 |
+
| support_gap_at_8 | 144 | 0.4833 | [0.3650, 0.6041] | 0.7508 | 0.5202 |
|
| 31 |
+
| total_car_to_hidden_at_8 | 144 | 0.9698 | [0.8262, 1.1140] | 1.2607 | 1.0425 |
|
workspace/runs/ctt_test_rollout_comparison/measured_metrics/table.tex
CHANGED
|
@@ -3,26 +3,30 @@
|
|
| 3 |
\toprule
|
| 4 |
Metric & N & Micro mean & CI low & CI high \\
|
| 5 |
\midrule
|
|
|
|
| 6 |
base\_success & 144 & 0.2847 & 0.2153 & 0.3542 \\
|
| 7 |
-
base\_utility & 144 & 0.7477 & 0.
|
| 8 |
-
branch\_car\_at\_8 & 144 & 0.5605 & 0.
|
| 9 |
-
|
| 10 |
-
hidden\_chart\_oracle\
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
proposal\_oracle\
|
| 15 |
-
proposal\_oracle\
|
| 16 |
-
proposal\_oracle\
|
|
|
|
|
|
|
|
|
|
| 17 |
selected\_success\_at\_8 & 144 & 0.2222 & 0.1528 & 0.2917 \\
|
| 18 |
-
selected\_success\_gain\_over\_base\_at\_8 & 144 & -0.0625 & -0.
|
| 19 |
-
selected\_utility\_at\_8 & 144 & 0.5485 & 0.
|
| 20 |
-
selected\_utility\_gain\_over\_base\_at\_8 & 144 & -0.1992 & -0.3477 & -0.
|
| 21 |
-
selector\_regret\_at\_8 & 144 & 0.5605 & 0.
|
| 22 |
-
success\_selector\_gap\_at\_8 & 144 & 0.2847 & 0.
|
| 23 |
success\_support\_gap\_at\_8 & 144 & 0.2639 & 0.1944 & 0.3333 \\
|
| 24 |
success\_total\_car\_to\_hidden\_at\_8 & 144 & 0.5069 & 0.4236 & 0.5903 \\
|
| 25 |
-
support\_gap\_at\_8 & 144 & 0.4833 & 0.
|
| 26 |
-
total\_car\_to\_hidden\_at\_8 & 144 & 0.9698 & 0.
|
| 27 |
\bottomrule
|
| 28 |
\end{tabular}
|
|
|
|
| 3 |
\toprule
|
| 4 |
Metric & N & Micro mean & CI low & CI high \\
|
| 5 |
\midrule
|
| 6 |
+
base\_safety\_label\_known & 144 & 0.0000 & 0.0000 & 0.0000 \\
|
| 7 |
base\_success & 144 & 0.2847 & 0.2153 & 0.3542 \\
|
| 8 |
+
base\_utility & 144 & 0.7477 & 0.6173 & 0.8788 \\
|
| 9 |
+
branch\_car\_at\_8 & 144 & 0.5605 & 0.4345 & 0.7065 \\
|
| 10 |
+
generated\_safety\_label\_coverage\_at\_8 & 144 & 0.0000 & 0.0000 & 0.0000 \\
|
| 11 |
+
hidden\_chart\_oracle\_success\_at\_8 & 144 & 0.7292 & 0.6597 & 0.7986 \\
|
| 12 |
+
hidden\_chart\_oracle\_utility\_at\_8 & 144 & 1.5183 & 1.3941 & 1.6438 \\
|
| 13 |
+
outcome\_ptr\_at\_8 & 144 & 0.5278 & 0.4514 & 0.6042 \\
|
| 14 |
+
pairwise\_causal\_calibration\_ece & 144 & 0.1730 & 0.1547 & 0.1911 \\
|
| 15 |
+
proposal\_oracle\_safety\_label\_known\_at\_8 & 144 & 0.0000 & 0.0000 & 0.0000 \\
|
| 16 |
+
proposal\_oracle\_success\_at\_8 & 144 & 0.5069 & 0.4306 & 0.5903 \\
|
| 17 |
+
proposal\_oracle\_success\_gain\_over\_base\_at\_8 & 144 & 0.2222 & 0.1319 & 0.3056 \\
|
| 18 |
+
proposal\_oracle\_utility\_at\_8 & 144 & 1.1090 & 0.9685 & 1.2662 \\
|
| 19 |
+
proposal\_oracle\_utility\_gain\_over\_base\_at\_8 & 144 & 0.3613 & 0.2109 & 0.5091 \\
|
| 20 |
+
selected\_safety\_label\_known\_at\_8 & 144 & 0.0000 & 0.0000 & 0.0000 \\
|
| 21 |
selected\_success\_at\_8 & 144 & 0.2222 & 0.1528 & 0.2917 \\
|
| 22 |
+
selected\_success\_gain\_over\_base\_at\_8 & 144 & -0.0625 & -0.1458 & 0.0139 \\
|
| 23 |
+
selected\_utility\_at\_8 & 144 & 0.5485 & 0.4193 & 0.6933 \\
|
| 24 |
+
selected\_utility\_gain\_over\_base\_at\_8 & 144 & -0.1992 & -0.3477 & -0.0505 \\
|
| 25 |
+
selector\_regret\_at\_8 & 144 & 0.5605 & 0.4345 & 0.7065 \\
|
| 26 |
+
success\_selector\_gap\_at\_8 & 144 & 0.2847 & 0.2153 & 0.3611 \\
|
| 27 |
success\_support\_gap\_at\_8 & 144 & 0.2639 & 0.1944 & 0.3333 \\
|
| 28 |
success\_total\_car\_to\_hidden\_at\_8 & 144 & 0.5069 & 0.4236 & 0.5903 \\
|
| 29 |
+
support\_gap\_at\_8 & 144 & 0.4833 & 0.3650 & 0.6041 \\
|
| 30 |
+
total\_car\_to\_hidden\_at\_8 & 144 & 0.9698 & 0.8262 & 1.1140 \\
|
| 31 |
\bottomrule
|
| 32 |
\end{tabular}
|
workspace/runs/ctt_test_rollout_comparison/metrics.json
CHANGED
|
@@ -9,30 +9,202 @@
|
|
| 9 |
],
|
| 10 |
"source_content_hash": "90b431047a1f35cec095cbafa7de5bc0126cb0d0f4db3ff31843dca369977957",
|
| 11 |
"success_summary": {
|
|
|
|
|
|
|
| 12 |
"base_success_rate": 0.2847222222222222,
|
|
|
|
| 13 |
"base_utility_mean": 0.7477176273686281,
|
|
|
|
|
|
|
| 14 |
"hidden_chart_oracle_success_rate": 0.7291666666666666,
|
| 15 |
"hidden_chart_oracle_utility_mean": 1.5183161981403828,
|
| 16 |
"max_restore_error": 1.4901161193847656e-08,
|
|
|
|
| 17 |
"proposal_oracle_success_gain_over_base": 0.2222222222222222,
|
| 18 |
"proposal_oracle_success_rate": 0.5069444444444444,
|
|
|
|
| 19 |
"proposal_oracle_utility_mean": 1.109000739677059,
|
|
|
|
| 20 |
"selected_success_gain_over_base": -0.0625,
|
| 21 |
"selected_success_rate": 0.2222222222222222,
|
|
|
|
| 22 |
"selected_utility_mean": 0.5485236065190596,
|
| 23 |
"success_selector_gap": 0.2847222222222222,
|
| 24 |
"success_support_gap": 0.2638888888888889
|
| 25 |
},
|
| 26 |
"summary": {
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
"base_success": {
|
| 28 |
"macro_by_seed": {
|
| 29 |
"ci": {
|
| 30 |
"confidence": 0.95,
|
| 31 |
-
"high": 0.
|
| 32 |
-
"low": 0.
|
| 33 |
"mean": 0.2916666666666667,
|
| 34 |
"n": 120,
|
| 35 |
-
"num_samples":
|
| 36 |
},
|
| 37 |
"mean": 0.2916666666666667,
|
| 38 |
"per_seed": {
|
|
@@ -161,11 +333,11 @@
|
|
| 161 |
"macro_by_task": {
|
| 162 |
"ci": {
|
| 163 |
"confidence": 0.95,
|
| 164 |
-
"high": 0.
|
| 165 |
"low": 0.0808080808080808,
|
| 166 |
"mean": 0.1954747474747475,
|
| 167 |
"n": 5,
|
| 168 |
-
"num_samples":
|
| 169 |
},
|
| 170 |
"mean": 0.1954747474747475,
|
| 171 |
"per_task": {
|
|
@@ -183,7 +355,7 @@
|
|
| 183 |
"low": 0.2152777777777778,
|
| 184 |
"mean": 0.2847222222222222,
|
| 185 |
"n": 144,
|
| 186 |
-
"num_samples":
|
| 187 |
},
|
| 188 |
"n": 144
|
| 189 |
},
|
|
@@ -191,11 +363,11 @@
|
|
| 191 |
"macro_by_seed": {
|
| 192 |
"ci": {
|
| 193 |
"confidence": 0.95,
|
| 194 |
-
"high": 0.
|
| 195 |
-
"low": 0.
|
| 196 |
"mean": 0.7857063563108873,
|
| 197 |
"n": 120,
|
| 198 |
-
"num_samples":
|
| 199 |
},
|
| 200 |
"mean": 0.7857063563108873,
|
| 201 |
"per_seed": {
|
|
@@ -324,11 +496,11 @@
|
|
| 324 |
"macro_by_task": {
|
| 325 |
"ci": {
|
| 326 |
"confidence": 0.95,
|
| 327 |
-
"high": 0.
|
| 328 |
-
"low": 0.
|
| 329 |
"mean": 0.7040124819563852,
|
| 330 |
"n": 5,
|
| 331 |
-
"num_samples":
|
| 332 |
},
|
| 333 |
"mean": 0.7040124819563852,
|
| 334 |
"per_task": {
|
|
@@ -342,11 +514,11 @@
|
|
| 342 |
"metric": "base_utility",
|
| 343 |
"micro": {
|
| 344 |
"confidence": 0.95,
|
| 345 |
-
"high": 0.
|
| 346 |
-
"low": 0.
|
| 347 |
"mean": 0.7477176273686281,
|
| 348 |
"n": 144,
|
| 349 |
-
"num_samples":
|
| 350 |
},
|
| 351 |
"n": 144
|
| 352 |
},
|
|
@@ -354,11 +526,11 @@
|
|
| 354 |
"macro_by_seed": {
|
| 355 |
"ci": {
|
| 356 |
"confidence": 0.95,
|
| 357 |
-
"high": 0.
|
| 358 |
-
"low": 0.
|
| 359 |
"mean": 0.5889192757033419,
|
| 360 |
"n": 120,
|
| 361 |
-
"num_samples":
|
| 362 |
},
|
| 363 |
"mean": 0.5889192757033419,
|
| 364 |
"per_seed": {
|
|
@@ -487,11 +659,11 @@
|
|
| 487 |
"macro_by_task": {
|
| 488 |
"ci": {
|
| 489 |
"confidence": 0.95,
|
| 490 |
-
"high": 0.
|
| 491 |
-
"low": 0.
|
| 492 |
"mean": 0.5837108520707774,
|
| 493 |
"n": 5,
|
| 494 |
-
"num_samples":
|
| 495 |
},
|
| 496 |
"mean": 0.5837108520707774,
|
| 497 |
"per_task": {
|
|
@@ -505,11 +677,174 @@
|
|
| 505 |
"metric": "branch_car_at_8",
|
| 506 |
"micro": {
|
| 507 |
"confidence": 0.95,
|
| 508 |
-
"high": 0.
|
| 509 |
-
"low": 0.
|
| 510 |
"mean": 0.5604771331579994,
|
| 511 |
"n": 144,
|
| 512 |
-
"num_samples":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 513 |
},
|
| 514 |
"n": 144
|
| 515 |
},
|
|
@@ -518,10 +853,10 @@
|
|
| 518 |
"ci": {
|
| 519 |
"confidence": 0.95,
|
| 520 |
"high": 0.8555555555555556,
|
| 521 |
-
"low": 0.
|
| 522 |
"mean": 0.7854166666666668,
|
| 523 |
"n": 120,
|
| 524 |
-
"num_samples":
|
| 525 |
},
|
| 526 |
"mean": 0.7854166666666668,
|
| 527 |
"per_seed": {
|
|
@@ -654,7 +989,7 @@
|
|
| 654 |
"low": 0.6786666666666668,
|
| 655 |
"mean": 0.8706666666666667,
|
| 656 |
"n": 5,
|
| 657 |
-
"num_samples":
|
| 658 |
},
|
| 659 |
"mean": 0.8706666666666667,
|
| 660 |
"per_task": {
|
|
@@ -668,11 +1003,11 @@
|
|
| 668 |
"metric": "hidden_chart_oracle_success_at_8",
|
| 669 |
"micro": {
|
| 670 |
"confidence": 0.95,
|
| 671 |
-
"high": 0.
|
| 672 |
"low": 0.6597222222222222,
|
| 673 |
"mean": 0.7291666666666666,
|
| 674 |
"n": 144,
|
| 675 |
-
"num_samples":
|
| 676 |
},
|
| 677 |
"n": 144
|
| 678 |
},
|
|
@@ -680,11 +1015,11 @@
|
|
| 680 |
"macro_by_seed": {
|
| 681 |
"ci": {
|
| 682 |
"confidence": 0.95,
|
| 683 |
-
"high": 1.
|
| 684 |
-
"low": 1.
|
| 685 |
"mean": 1.6190625454609595,
|
| 686 |
"n": 120,
|
| 687 |
-
"num_samples":
|
| 688 |
},
|
| 689 |
"mean": 1.6190625454609595,
|
| 690 |
"per_seed": {
|
|
@@ -817,7 +1152,7 @@
|
|
| 817 |
"low": 1.428388743619124,
|
| 818 |
"mean": 1.7696735892097153,
|
| 819 |
"n": 5,
|
| 820 |
-
"num_samples":
|
| 821 |
},
|
| 822 |
"mean": 1.7696735892097153,
|
| 823 |
"per_task": {
|
|
@@ -831,11 +1166,11 @@
|
|
| 831 |
"metric": "hidden_chart_oracle_utility_at_8",
|
| 832 |
"micro": {
|
| 833 |
"confidence": 0.95,
|
| 834 |
-
"high": 1.
|
| 835 |
-
"low": 1.
|
| 836 |
"mean": 1.5183161981403828,
|
| 837 |
"n": 144,
|
| 838 |
-
"num_samples":
|
| 839 |
},
|
| 840 |
"n": 144
|
| 841 |
},
|
|
@@ -843,11 +1178,11 @@
|
|
| 843 |
"macro_by_seed": {
|
| 844 |
"ci": {
|
| 845 |
"confidence": 0.95,
|
| 846 |
-
"high": 0.
|
| 847 |
"low": 0.43680555555555556,
|
| 848 |
"mean": 0.5243055555555556,
|
| 849 |
"n": 120,
|
| 850 |
-
"num_samples":
|
| 851 |
},
|
| 852 |
"mean": 0.5243055555555556,
|
| 853 |
"per_seed": {
|
|
@@ -976,11 +1311,11 @@
|
|
| 976 |
"macro_by_task": {
|
| 977 |
"ci": {
|
| 978 |
"confidence": 0.95,
|
| 979 |
-
"high": 0.
|
| 980 |
-
"low": 0.
|
| 981 |
"mean": 0.5128888888888888,
|
| 982 |
"n": 5,
|
| 983 |
-
"num_samples":
|
| 984 |
},
|
| 985 |
"mean": 0.5128888888888888,
|
| 986 |
"per_task": {
|
|
@@ -995,10 +1330,10 @@
|
|
| 995 |
"micro": {
|
| 996 |
"confidence": 0.95,
|
| 997 |
"high": 0.6041666666666666,
|
| 998 |
-
"low": 0.
|
| 999 |
"mean": 0.5277777777777778,
|
| 1000 |
"n": 144,
|
| 1001 |
-
"num_samples":
|
| 1002 |
},
|
| 1003 |
"n": 144
|
| 1004 |
},
|
|
@@ -1006,11 +1341,11 @@
|
|
| 1006 |
"macro_by_seed": {
|
| 1007 |
"ci": {
|
| 1008 |
"confidence": 0.95,
|
| 1009 |
-
"high": 0.
|
| 1010 |
-
"low": 0.
|
| 1011 |
"mean": 0.17180047070386753,
|
| 1012 |
"n": 120,
|
| 1013 |
-
"num_samples":
|
| 1014 |
},
|
| 1015 |
"mean": 0.17180047070386753,
|
| 1016 |
"per_seed": {
|
|
@@ -1096,72 +1431,235 @@
|
|
| 1096 |
"412102563/train1": 0.05497825758025142,
|
| 1097 |
"412102563/train2": 0.23421550754433385,
|
| 1098 |
"440577711/train0": 0.0,
|
| 1099 |
-
"440577711/train1": 0.08649062167899924,
|
| 1100 |
-
"440577711/train2": 0.15795943889155128,
|
| 1101 |
-
"441046872/train0": 0.07559488024963246,
|
| 1102 |
-
"441046872/train1": 0.03055053741629537,
|
| 1103 |
-
"441046872/train2": 0.10573947390621682,
|
| 1104 |
-
"509430392/train0": 0.05599945856896821,
|
| 1105 |
-
"509430392/train1": 0.29083191353306787,
|
| 1106 |
-
"509430392/train2": 0.15055463357076268,
|
| 1107 |
-
"585904130/train0": 0.14940863096718615,
|
| 1108 |
-
"585904130/train1": 0.24258759953449954,
|
| 1109 |
-
"585904130/train2": 0.31468676305833865,
|
| 1110 |
-
"608016290/train0": 0.3387708317969876,
|
| 1111 |
-
"608016290/train1": 0.1955644997142938,
|
| 1112 |
-
"608016290/train2": 0.159510713855325,
|
| 1113 |
-
"626486813/train0": 0.21441221141580127,
|
| 1114 |
-
"626486813/train1": 0.170949273586752,
|
| 1115 |
-
"626486813/train2": 0.16604576952462607,
|
| 1116 |
-
"636680137/train0": 0.2814239671003126,
|
| 1117 |
-
"636680137/train1": 0.2625830101532915,
|
| 1118 |
-
"636680137/train2": 0.12020315308522647,
|
| 1119 |
-
"655884605/train0": 0.14628351854390587,
|
| 1120 |
-
"655884605/train1": 0.0690564909179267,
|
| 1121 |
-
"655884605/train2": 0.320973287648995,
|
| 1122 |
-
"672328262/train0": 0.16662194482040366,
|
| 1123 |
-
"672328262/train1": 0.18784609006259695,
|
| 1124 |
-
"672328262/train2": 0.2998508752953551,
|
| 1125 |
-
"701667486/train0": 0.13404567006107426,
|
| 1126 |
-
"701667486/train1": 0.12687721623280993,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1127 |
"701667486/train2": 0.0,
|
| 1128 |
-
"747260437/train0": 0.
|
| 1129 |
-
"747260437/train1": 0.
|
| 1130 |
-
"747260437/train2": 0.
|
| 1131 |
-
"750448296/train0": 0.
|
| 1132 |
-
"750448296/train1": 0.
|
| 1133 |
-
"750448296/train2": 0.
|
| 1134 |
-
"767747424/train0": 0.
|
| 1135 |
-
"767747424/train1": 0.
|
| 1136 |
-
"767747424/train2": 0.
|
| 1137 |
}
|
| 1138 |
},
|
| 1139 |
"macro_by_task": {
|
| 1140 |
"ci": {
|
| 1141 |
"confidence": 0.95,
|
| 1142 |
-
"high": 0.
|
| 1143 |
-
"low": 0.
|
| 1144 |
-
"mean": 0.
|
| 1145 |
"n": 5,
|
| 1146 |
-
"num_samples":
|
| 1147 |
},
|
| 1148 |
-
"mean": 0.
|
| 1149 |
"per_task": {
|
| 1150 |
-
"LiftPegUpright-v1": 0.
|
| 1151 |
-
"PickCube-v1": 0.
|
| 1152 |
-
"PullCube-v1": 0.
|
| 1153 |
-
"PushCube-v1": 0.
|
| 1154 |
-
"StackCube-v1": 0.
|
| 1155 |
}
|
| 1156 |
},
|
| 1157 |
-
"metric": "
|
| 1158 |
"micro": {
|
| 1159 |
"confidence": 0.95,
|
| 1160 |
-
"high": 0.
|
| 1161 |
-
"low": 0.
|
| 1162 |
-
"mean": 0.
|
| 1163 |
"n": 144,
|
| 1164 |
-
"num_samples":
|
| 1165 |
},
|
| 1166 |
"n": 144
|
| 1167 |
},
|
|
@@ -1169,11 +1667,11 @@
|
|
| 1169 |
"macro_by_seed": {
|
| 1170 |
"ci": {
|
| 1171 |
"confidence": 0.95,
|
| 1172 |
-
"high": 0.
|
| 1173 |
-
"low": 0.
|
| 1174 |
"mean": 0.53125,
|
| 1175 |
"n": 120,
|
| 1176 |
-
"num_samples":
|
| 1177 |
},
|
| 1178 |
"mean": 0.53125,
|
| 1179 |
"per_seed": {
|
|
@@ -1302,11 +1800,11 @@
|
|
| 1302 |
"macro_by_task": {
|
| 1303 |
"ci": {
|
| 1304 |
"confidence": 0.95,
|
| 1305 |
-
"high": 0.
|
| 1306 |
"low": 0.21818181818181817,
|
| 1307 |
"mean": 0.4755151515151515,
|
| 1308 |
"n": 5,
|
| 1309 |
-
"num_samples":
|
| 1310 |
},
|
| 1311 |
"mean": 0.4755151515151515,
|
| 1312 |
"per_task": {
|
|
@@ -1320,11 +1818,11 @@
|
|
| 1320 |
"metric": "proposal_oracle_success_at_8",
|
| 1321 |
"micro": {
|
| 1322 |
"confidence": 0.95,
|
| 1323 |
-
"high": 0.
|
| 1324 |
"low": 0.4305555555555556,
|
| 1325 |
"mean": 0.5069444444444444,
|
| 1326 |
"n": 144,
|
| 1327 |
-
"num_samples":
|
| 1328 |
},
|
| 1329 |
"n": 144
|
| 1330 |
},
|
|
@@ -1332,11 +1830,11 @@
|
|
| 1332 |
"macro_by_seed": {
|
| 1333 |
"ci": {
|
| 1334 |
"confidence": 0.95,
|
| 1335 |
-
"high": 0.
|
| 1336 |
"low": 0.14513888888888887,
|
| 1337 |
"mean": 0.23958333333333334,
|
| 1338 |
"n": 120,
|
| 1339 |
-
"num_samples":
|
| 1340 |
},
|
| 1341 |
"mean": 0.23958333333333334,
|
| 1342 |
"per_seed": {
|
|
@@ -1466,10 +1964,10 @@
|
|
| 1466 |
"ci": {
|
| 1467 |
"confidence": 0.95,
|
| 1468 |
"high": 0.4760000000000001,
|
| 1469 |
-
"low": 0.
|
| 1470 |
"mean": 0.28004040404040403,
|
| 1471 |
"n": 5,
|
| 1472 |
-
"num_samples":
|
| 1473 |
},
|
| 1474 |
"mean": 0.28004040404040403,
|
| 1475 |
"per_task": {
|
|
@@ -1483,11 +1981,11 @@
|
|
| 1483 |
"metric": "proposal_oracle_success_gain_over_base_at_8",
|
| 1484 |
"micro": {
|
| 1485 |
"confidence": 0.95,
|
| 1486 |
-
"high": 0.
|
| 1487 |
"low": 0.13194444444444445,
|
| 1488 |
"mean": 0.2222222222222222,
|
| 1489 |
"n": 144,
|
| 1490 |
-
"num_samples":
|
| 1491 |
},
|
| 1492 |
"n": 144
|
| 1493 |
},
|
|
@@ -1495,11 +1993,11 @@
|
|
| 1495 |
"macro_by_seed": {
|
| 1496 |
"ci": {
|
| 1497 |
"confidence": 0.95,
|
| 1498 |
-
"high": 1.
|
| 1499 |
-
"low": 1.
|
| 1500 |
"mean": 1.1654707208521562,
|
| 1501 |
"n": 120,
|
| 1502 |
-
"num_samples":
|
| 1503 |
},
|
| 1504 |
"mean": 1.1654707208521562,
|
| 1505 |
"per_seed": {
|
|
@@ -1628,11 +2126,11 @@
|
|
| 1628 |
"macro_by_task": {
|
| 1629 |
"ci": {
|
| 1630 |
"confidence": 0.95,
|
| 1631 |
-
"high": 1.
|
| 1632 |
"low": 0.6483407714106143,
|
| 1633 |
"mean": 1.0926857944947828,
|
| 1634 |
"n": 5,
|
| 1635 |
-
"num_samples":
|
| 1636 |
},
|
| 1637 |
"mean": 1.0926857944947828,
|
| 1638 |
"per_task": {
|
|
@@ -1646,11 +2144,11 @@
|
|
| 1646 |
"metric": "proposal_oracle_utility_at_8",
|
| 1647 |
"micro": {
|
| 1648 |
"confidence": 0.95,
|
| 1649 |
-
"high": 1.
|
| 1650 |
-
"low": 0.
|
| 1651 |
"mean": 1.109000739677059,
|
| 1652 |
"n": 144,
|
| 1653 |
-
"num_samples":
|
| 1654 |
},
|
| 1655 |
"n": 144
|
| 1656 |
},
|
|
@@ -1658,11 +2156,11 @@
|
|
| 1658 |
"macro_by_seed": {
|
| 1659 |
"ci": {
|
| 1660 |
"confidence": 0.95,
|
| 1661 |
-
"high": 0.
|
| 1662 |
-
"low": 0.
|
| 1663 |
"mean": 0.37976436454126894,
|
| 1664 |
"n": 120,
|
| 1665 |
-
"num_samples":
|
| 1666 |
},
|
| 1667 |
"mean": 0.37976436454126894,
|
| 1668 |
"per_seed": {
|
|
@@ -1791,11 +2289,11 @@
|
|
| 1791 |
"macro_by_task": {
|
| 1792 |
"ci": {
|
| 1793 |
"confidence": 0.95,
|
| 1794 |
-
"high": 0.
|
| 1795 |
-
"low": -0.
|
| 1796 |
"mean": 0.3886733125383975,
|
| 1797 |
"n": 5,
|
| 1798 |
-
"num_samples":
|
| 1799 |
},
|
| 1800 |
"mean": 0.3886733125383975,
|
| 1801 |
"per_task": {
|
|
@@ -1809,11 +2307,174 @@
|
|
| 1809 |
"metric": "proposal_oracle_utility_gain_over_base_at_8",
|
| 1810 |
"micro": {
|
| 1811 |
"confidence": 0.95,
|
| 1812 |
-
"high": 0.
|
| 1813 |
-
"low": 0.
|
| 1814 |
"mean": 0.36128311230843085,
|
| 1815 |
"n": 144,
|
| 1816 |
-
"num_samples":
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1817 |
},
|
| 1818 |
"n": 144
|
| 1819 |
},
|
|
@@ -1822,10 +2483,10 @@
|
|
| 1822 |
"ci": {
|
| 1823 |
"confidence": 0.95,
|
| 1824 |
"high": 0.3055555555555555,
|
| 1825 |
-
"low": 0.
|
| 1826 |
"mean": 0.23055555555555554,
|
| 1827 |
"n": 120,
|
| 1828 |
-
"num_samples":
|
| 1829 |
},
|
| 1830 |
"mean": 0.23055555555555554,
|
| 1831 |
"per_seed": {
|
|
@@ -1958,7 +2619,7 @@
|
|
| 1958 |
"low": 0.07490909090909091,
|
| 1959 |
"mean": 0.18602020202020203,
|
| 1960 |
"n": 5,
|
| 1961 |
-
"num_samples":
|
| 1962 |
},
|
| 1963 |
"mean": 0.18602020202020203,
|
| 1964 |
"per_task": {
|
|
@@ -1976,7 +2637,7 @@
|
|
| 1976 |
"low": 0.1527777777777778,
|
| 1977 |
"mean": 0.2222222222222222,
|
| 1978 |
"n": 144,
|
| 1979 |
-
"num_samples":
|
| 1980 |
},
|
| 1981 |
"n": 144
|
| 1982 |
},
|
|
@@ -1984,11 +2645,11 @@
|
|
| 1984 |
"macro_by_seed": {
|
| 1985 |
"ci": {
|
| 1986 |
"confidence": 0.95,
|
| 1987 |
-
"high": 0.
|
| 1988 |
-
"low": -0.
|
| 1989 |
"mean": -0.061111111111111116,
|
| 1990 |
"n": 120,
|
| 1991 |
-
"num_samples":
|
| 1992 |
},
|
| 1993 |
"mean": -0.061111111111111116,
|
| 1994 |
"per_seed": {
|
|
@@ -2117,11 +2778,11 @@
|
|
| 2117 |
"macro_by_task": {
|
| 2118 |
"ci": {
|
| 2119 |
"confidence": 0.95,
|
| 2120 |
-
"high": 0.
|
| 2121 |
"low": -0.08412121212121212,
|
| 2122 |
"mean": -0.009454545454545455,
|
| 2123 |
"n": 5,
|
| 2124 |
-
"num_samples":
|
| 2125 |
},
|
| 2126 |
"mean": -0.009454545454545455,
|
| 2127 |
"per_task": {
|
|
@@ -2136,10 +2797,10 @@
|
|
| 2136 |
"micro": {
|
| 2137 |
"confidence": 0.95,
|
| 2138 |
"high": 0.013888888888888888,
|
| 2139 |
-
"low": -0.
|
| 2140 |
"mean": -0.0625,
|
| 2141 |
"n": 144,
|
| 2142 |
-
"num_samples":
|
| 2143 |
},
|
| 2144 |
"n": 144
|
| 2145 |
},
|
|
@@ -2147,11 +2808,11 @@
|
|
| 2147 |
"macro_by_seed": {
|
| 2148 |
"ci": {
|
| 2149 |
"confidence": 0.95,
|
| 2150 |
-
"high": 0.
|
| 2151 |
-
"low": 0.
|
| 2152 |
"mean": 0.5765514451488142,
|
| 2153 |
"n": 120,
|
| 2154 |
-
"num_samples":
|
| 2155 |
},
|
| 2156 |
"mean": 0.5765514451488142,
|
| 2157 |
"per_seed": {
|
|
@@ -2281,10 +2942,10 @@
|
|
| 2281 |
"ci": {
|
| 2282 |
"confidence": 0.95,
|
| 2283 |
"high": 0.7802921726552374,
|
| 2284 |
-
"low": 0.
|
| 2285 |
"mean": 0.5089749424240053,
|
| 2286 |
"n": 5,
|
| 2287 |
-
"num_samples":
|
| 2288 |
},
|
| 2289 |
"mean": 0.5089749424240053,
|
| 2290 |
"per_task": {
|
|
@@ -2298,11 +2959,11 @@
|
|
| 2298 |
"metric": "selected_utility_at_8",
|
| 2299 |
"micro": {
|
| 2300 |
"confidence": 0.95,
|
| 2301 |
-
"high": 0.
|
| 2302 |
-
"low": 0.
|
| 2303 |
"mean": 0.5485236065190596,
|
| 2304 |
"n": 144,
|
| 2305 |
-
"num_samples":
|
| 2306 |
},
|
| 2307 |
"n": 144
|
| 2308 |
},
|
|
@@ -2310,11 +2971,11 @@
|
|
| 2310 |
"macro_by_seed": {
|
| 2311 |
"ci": {
|
| 2312 |
"confidence": 0.95,
|
| 2313 |
-
"high": -0.
|
| 2314 |
-
"low": -0.
|
| 2315 |
"mean": -0.20915491116207302,
|
| 2316 |
"n": 120,
|
| 2317 |
-
"num_samples":
|
| 2318 |
},
|
| 2319 |
"mean": -0.20915491116207302,
|
| 2320 |
"per_seed": {
|
|
@@ -2443,11 +3104,11 @@
|
|
| 2443 |
"macro_by_task": {
|
| 2444 |
"ci": {
|
| 2445 |
"confidence": 0.95,
|
| 2446 |
-
"high": -0.
|
| 2447 |
"low": -0.3532214832519548,
|
| 2448 |
"mean": -0.19503753953237987,
|
| 2449 |
"n": 5,
|
| 2450 |
-
"num_samples":
|
| 2451 |
},
|
| 2452 |
"mean": -0.19503753953237987,
|
| 2453 |
"per_task": {
|
|
@@ -2461,11 +3122,11 @@
|
|
| 2461 |
"metric": "selected_utility_gain_over_base_at_8",
|
| 2462 |
"micro": {
|
| 2463 |
"confidence": 0.95,
|
| 2464 |
-
"high": -0.
|
| 2465 |
"low": -0.34769073229431846,
|
| 2466 |
"mean": -0.1991940208495685,
|
| 2467 |
"n": 144,
|
| 2468 |
-
"num_samples":
|
| 2469 |
},
|
| 2470 |
"n": 144
|
| 2471 |
},
|
|
@@ -2473,11 +3134,11 @@
|
|
| 2473 |
"macro_by_seed": {
|
| 2474 |
"ci": {
|
| 2475 |
"confidence": 0.95,
|
| 2476 |
-
"high": 0.
|
| 2477 |
-
"low": 0.
|
| 2478 |
"mean": 0.5889192757033419,
|
| 2479 |
"n": 120,
|
| 2480 |
-
"num_samples":
|
| 2481 |
},
|
| 2482 |
"mean": 0.5889192757033419,
|
| 2483 |
"per_seed": {
|
|
@@ -2606,11 +3267,11 @@
|
|
| 2606 |
"macro_by_task": {
|
| 2607 |
"ci": {
|
| 2608 |
"confidence": 0.95,
|
| 2609 |
-
"high": 0.
|
| 2610 |
-
"low": 0.
|
| 2611 |
"mean": 0.5837108520707774,
|
| 2612 |
"n": 5,
|
| 2613 |
-
"num_samples":
|
| 2614 |
},
|
| 2615 |
"mean": 0.5837108520707774,
|
| 2616 |
"per_task": {
|
|
@@ -2624,11 +3285,11 @@
|
|
| 2624 |
"metric": "selector_regret_at_8",
|
| 2625 |
"micro": {
|
| 2626 |
"confidence": 0.95,
|
| 2627 |
-
"high": 0.
|
| 2628 |
-
"low": 0.
|
| 2629 |
"mean": 0.5604771331579994,
|
| 2630 |
"n": 144,
|
| 2631 |
-
"num_samples":
|
| 2632 |
},
|
| 2633 |
"n": 144
|
| 2634 |
},
|
|
@@ -2636,11 +3297,11 @@
|
|
| 2636 |
"macro_by_seed": {
|
| 2637 |
"ci": {
|
| 2638 |
"confidence": 0.95,
|
| 2639 |
-
"high": 0.
|
| 2640 |
-
"low": 0.
|
| 2641 |
"mean": 0.3006944444444445,
|
| 2642 |
"n": 120,
|
| 2643 |
-
"num_samples":
|
| 2644 |
},
|
| 2645 |
"mean": 0.3006944444444445,
|
| 2646 |
"per_seed": {
|
|
@@ -2769,11 +3430,11 @@
|
|
| 2769 |
"macro_by_task": {
|
| 2770 |
"ci": {
|
| 2771 |
"confidence": 0.95,
|
| 2772 |
-
"high": 0.
|
| 2773 |
-
"low": 0.
|
| 2774 |
"mean": 0.28949494949494947,
|
| 2775 |
"n": 5,
|
| 2776 |
-
"num_samples":
|
| 2777 |
},
|
| 2778 |
"mean": 0.28949494949494947,
|
| 2779 |
"per_task": {
|
|
@@ -2788,10 +3449,10 @@
|
|
| 2788 |
"micro": {
|
| 2789 |
"confidence": 0.95,
|
| 2790 |
"high": 0.3611111111111111,
|
| 2791 |
-
"low": 0.
|
| 2792 |
"mean": 0.2847222222222222,
|
| 2793 |
"n": 144,
|
| 2794 |
-
"num_samples":
|
| 2795 |
},
|
| 2796 |
"n": 144
|
| 2797 |
},
|
|
@@ -2799,11 +3460,11 @@
|
|
| 2799 |
"macro_by_seed": {
|
| 2800 |
"ci": {
|
| 2801 |
"confidence": 0.95,
|
| 2802 |
-
"high": 0.
|
| 2803 |
"low": 0.21111111111111114,
|
| 2804 |
"mean": 0.2916666666666667,
|
| 2805 |
"n": 120,
|
| 2806 |
-
"num_samples":
|
| 2807 |
},
|
| 2808 |
"mean": 0.2916666666666667,
|
| 2809 |
"per_seed": {
|
|
@@ -2936,7 +3597,7 @@
|
|
| 2936 |
"low": 0.176,
|
| 2937 |
"mean": 0.43648484848484853,
|
| 2938 |
"n": 5,
|
| 2939 |
-
"num_samples":
|
| 2940 |
},
|
| 2941 |
"mean": 0.43648484848484853,
|
| 2942 |
"per_task": {
|
|
@@ -2954,7 +3615,7 @@
|
|
| 2954 |
"low": 0.19444444444444445,
|
| 2955 |
"mean": 0.2638888888888889,
|
| 2956 |
"n": 144,
|
| 2957 |
-
"num_samples":
|
| 2958 |
},
|
| 2959 |
"n": 144
|
| 2960 |
},
|
|
@@ -2962,11 +3623,11 @@
|
|
| 2962 |
"macro_by_seed": {
|
| 2963 |
"ci": {
|
| 2964 |
"confidence": 0.95,
|
| 2965 |
-
"high": 0.
|
| 2966 |
-
"low": 0.
|
| 2967 |
"mean": 0.554861111111111,
|
| 2968 |
"n": 120,
|
| 2969 |
-
"num_samples":
|
| 2970 |
},
|
| 2971 |
"mean": 0.554861111111111,
|
| 2972 |
"per_seed": {
|
|
@@ -3099,7 +3760,7 @@
|
|
| 3099 |
"low": 0.457979797979798,
|
| 3100 |
"mean": 0.6846464646464646,
|
| 3101 |
"n": 5,
|
| 3102 |
-
"num_samples":
|
| 3103 |
},
|
| 3104 |
"mean": 0.6846464646464646,
|
| 3105 |
"per_task": {
|
|
@@ -3117,7 +3778,7 @@
|
|
| 3117 |
"low": 0.4236111111111111,
|
| 3118 |
"mean": 0.5069444444444444,
|
| 3119 |
"n": 144,
|
| 3120 |
-
"num_samples":
|
| 3121 |
},
|
| 3122 |
"n": 144
|
| 3123 |
},
|
|
@@ -3125,11 +3786,11 @@
|
|
| 3125 |
"macro_by_seed": {
|
| 3126 |
"ci": {
|
| 3127 |
"confidence": 0.95,
|
| 3128 |
-
"high": 0.
|
| 3129 |
-
"low": 0.
|
| 3130 |
"mean": 0.5202379634047651,
|
| 3131 |
"n": 120,
|
| 3132 |
-
"num_samples":
|
| 3133 |
},
|
| 3134 |
"mean": 0.5202379634047651,
|
| 3135 |
"per_seed": {
|
|
@@ -3258,11 +3919,11 @@
|
|
| 3258 |
"macro_by_task": {
|
| 3259 |
"ci": {
|
| 3260 |
"confidence": 0.95,
|
| 3261 |
-
"high": 1.
|
| 3262 |
"low": 0.32555825475768907,
|
| 3263 |
"mean": 0.7507728851216686,
|
| 3264 |
"n": 5,
|
| 3265 |
-
"num_samples":
|
| 3266 |
},
|
| 3267 |
"mean": 0.7507728851216686,
|
| 3268 |
"per_task": {
|
|
@@ -3276,11 +3937,11 @@
|
|
| 3276 |
"metric": "support_gap_at_8",
|
| 3277 |
"micro": {
|
| 3278 |
"confidence": 0.95,
|
| 3279 |
-
"high": 0.
|
| 3280 |
-
"low": 0.
|
| 3281 |
"mean": 0.48333957182411624,
|
| 3282 |
"n": 144,
|
| 3283 |
-
"num_samples":
|
| 3284 |
},
|
| 3285 |
"n": 144
|
| 3286 |
},
|
|
@@ -3288,11 +3949,11 @@
|
|
| 3288 |
"macro_by_seed": {
|
| 3289 |
"ci": {
|
| 3290 |
"confidence": 0.95,
|
| 3291 |
-
"high": 1.
|
| 3292 |
-
"low": 0.
|
| 3293 |
"mean": 1.0425111003121452,
|
| 3294 |
"n": 120,
|
| 3295 |
-
"num_samples":
|
| 3296 |
},
|
| 3297 |
"mean": 1.0425111003121452,
|
| 3298 |
"per_seed": {
|
|
@@ -3422,10 +4083,10 @@
|
|
| 3422 |
"ci": {
|
| 3423 |
"confidence": 0.95,
|
| 3424 |
"high": 1.6644046113617492,
|
| 3425 |
-
"low": 0.
|
| 3426 |
"mean": 1.2606986467857102,
|
| 3427 |
"n": 5,
|
| 3428 |
-
"num_samples":
|
| 3429 |
},
|
| 3430 |
"mean": 1.2606986467857102,
|
| 3431 |
"per_task": {
|
|
@@ -3440,10 +4101,10 @@
|
|
| 3440 |
"micro": {
|
| 3441 |
"confidence": 0.95,
|
| 3442 |
"high": 1.1140025374075373,
|
| 3443 |
-
"low": 0.
|
| 3444 |
"mean": 0.9697925916213231,
|
| 3445 |
"n": 144,
|
| 3446 |
-
"num_samples":
|
| 3447 |
},
|
| 3448 |
"n": 144
|
| 3449 |
}
|
|
|
|
| 9 |
],
|
| 10 |
"source_content_hash": "90b431047a1f35cec095cbafa7de5bc0126cb0d0f4db3ff31843dca369977957",
|
| 11 |
"success_summary": {
|
| 12 |
+
"any_generated_unsafe_known": null,
|
| 13 |
+
"base_safety_label_known_rate": 0.0,
|
| 14 |
"base_success_rate": 0.2847222222222222,
|
| 15 |
+
"base_unsafe_rate_known": null,
|
| 16 |
"base_utility_mean": 0.7477176273686281,
|
| 17 |
+
"generated_safety_label_coverage": 0.0,
|
| 18 |
+
"generated_unsafe_rate_known": null,
|
| 19 |
"hidden_chart_oracle_success_rate": 0.7291666666666666,
|
| 20 |
"hidden_chart_oracle_utility_mean": 1.5183161981403828,
|
| 21 |
"max_restore_error": 1.4901161193847656e-08,
|
| 22 |
+
"proposal_oracle_safety_label_known_rate": 0.0,
|
| 23 |
"proposal_oracle_success_gain_over_base": 0.2222222222222222,
|
| 24 |
"proposal_oracle_success_rate": 0.5069444444444444,
|
| 25 |
+
"proposal_oracle_unsafe_rate_known": null,
|
| 26 |
"proposal_oracle_utility_mean": 1.109000739677059,
|
| 27 |
+
"selected_safety_label_known_rate": 0.0,
|
| 28 |
"selected_success_gain_over_base": -0.0625,
|
| 29 |
"selected_success_rate": 0.2222222222222222,
|
| 30 |
+
"selected_unsafe_rate_known": null,
|
| 31 |
"selected_utility_mean": 0.5485236065190596,
|
| 32 |
"success_selector_gap": 0.2847222222222222,
|
| 33 |
"success_support_gap": 0.2638888888888889
|
| 34 |
},
|
| 35 |
"summary": {
|
| 36 |
+
"base_safety_label_known": {
|
| 37 |
+
"macro_by_seed": {
|
| 38 |
+
"ci": {
|
| 39 |
+
"confidence": 0.95,
|
| 40 |
+
"high": 0.0,
|
| 41 |
+
"low": 0.0,
|
| 42 |
+
"mean": 0.0,
|
| 43 |
+
"n": 120,
|
| 44 |
+
"num_samples": 1000
|
| 45 |
+
},
|
| 46 |
+
"mean": 0.0,
|
| 47 |
+
"per_seed": {
|
| 48 |
+
"1073161778/train0": 0.0,
|
| 49 |
+
"1073161778/train1": 0.0,
|
| 50 |
+
"1073161778/train2": 0.0,
|
| 51 |
+
"1113371261/train0": 0.0,
|
| 52 |
+
"1113371261/train1": 0.0,
|
| 53 |
+
"1113371261/train2": 0.0,
|
| 54 |
+
"1151410346/train0": 0.0,
|
| 55 |
+
"1151410346/train1": 0.0,
|
| 56 |
+
"1151410346/train2": 0.0,
|
| 57 |
+
"1261930456/train0": 0.0,
|
| 58 |
+
"1261930456/train1": 0.0,
|
| 59 |
+
"1261930456/train2": 0.0,
|
| 60 |
+
"1324133122/train0": 0.0,
|
| 61 |
+
"1324133122/train1": 0.0,
|
| 62 |
+
"1324133122/train2": 0.0,
|
| 63 |
+
"1324642411/train0": 0.0,
|
| 64 |
+
"1324642411/train1": 0.0,
|
| 65 |
+
"1324642411/train2": 0.0,
|
| 66 |
+
"1375492786/train0": 0.0,
|
| 67 |
+
"1375492786/train1": 0.0,
|
| 68 |
+
"1375492786/train2": 0.0,
|
| 69 |
+
"141984615/train0": 0.0,
|
| 70 |
+
"141984615/train1": 0.0,
|
| 71 |
+
"141984615/train2": 0.0,
|
| 72 |
+
"1461069713/train0": 0.0,
|
| 73 |
+
"1461069713/train1": 0.0,
|
| 74 |
+
"1461069713/train2": 0.0,
|
| 75 |
+
"1480856810/train0": 0.0,
|
| 76 |
+
"1480856810/train1": 0.0,
|
| 77 |
+
"1480856810/train2": 0.0,
|
| 78 |
+
"1509980664/train0": 0.0,
|
| 79 |
+
"1509980664/train1": 0.0,
|
| 80 |
+
"1509980664/train2": 0.0,
|
| 81 |
+
"1522272686/train0": 0.0,
|
| 82 |
+
"1522272686/train1": 0.0,
|
| 83 |
+
"1522272686/train2": 0.0,
|
| 84 |
+
"1549324805/train0": 0.0,
|
| 85 |
+
"1549324805/train1": 0.0,
|
| 86 |
+
"1549324805/train2": 0.0,
|
| 87 |
+
"1578990937/train0": 0.0,
|
| 88 |
+
"1578990937/train1": 0.0,
|
| 89 |
+
"1578990937/train2": 0.0,
|
| 90 |
+
"1747178030/train0": 0.0,
|
| 91 |
+
"1747178030/train1": 0.0,
|
| 92 |
+
"1747178030/train2": 0.0,
|
| 93 |
+
"1790188787/train0": 0.0,
|
| 94 |
+
"1790188787/train1": 0.0,
|
| 95 |
+
"1790188787/train2": 0.0,
|
| 96 |
+
"1822210471/train0": 0.0,
|
| 97 |
+
"1822210471/train1": 0.0,
|
| 98 |
+
"1822210471/train2": 0.0,
|
| 99 |
+
"1845489013/train0": 0.0,
|
| 100 |
+
"1845489013/train1": 0.0,
|
| 101 |
+
"1845489013/train2": 0.0,
|
| 102 |
+
"1999022574/train0": 0.0,
|
| 103 |
+
"1999022574/train1": 0.0,
|
| 104 |
+
"1999022574/train2": 0.0,
|
| 105 |
+
"2012211287/train0": 0.0,
|
| 106 |
+
"2012211287/train1": 0.0,
|
| 107 |
+
"2012211287/train2": 0.0,
|
| 108 |
+
"2038594191/train0": 0.0,
|
| 109 |
+
"2038594191/train1": 0.0,
|
| 110 |
+
"2038594191/train2": 0.0,
|
| 111 |
+
"2054899126/train0": 0.0,
|
| 112 |
+
"2054899126/train1": 0.0,
|
| 113 |
+
"2054899126/train2": 0.0,
|
| 114 |
+
"2061635888/train0": 0.0,
|
| 115 |
+
"2061635888/train1": 0.0,
|
| 116 |
+
"2061635888/train2": 0.0,
|
| 117 |
+
"211245842/train0": 0.0,
|
| 118 |
+
"211245842/train1": 0.0,
|
| 119 |
+
"211245842/train2": 0.0,
|
| 120 |
+
"278008524/train0": 0.0,
|
| 121 |
+
"278008524/train1": 0.0,
|
| 122 |
+
"278008524/train2": 0.0,
|
| 123 |
+
"307068879/train0": 0.0,
|
| 124 |
+
"307068879/train1": 0.0,
|
| 125 |
+
"307068879/train2": 0.0,
|
| 126 |
+
"412102563/train0": 0.0,
|
| 127 |
+
"412102563/train1": 0.0,
|
| 128 |
+
"412102563/train2": 0.0,
|
| 129 |
+
"440577711/train0": 0.0,
|
| 130 |
+
"440577711/train1": 0.0,
|
| 131 |
+
"440577711/train2": 0.0,
|
| 132 |
+
"441046872/train0": 0.0,
|
| 133 |
+
"441046872/train1": 0.0,
|
| 134 |
+
"441046872/train2": 0.0,
|
| 135 |
+
"509430392/train0": 0.0,
|
| 136 |
+
"509430392/train1": 0.0,
|
| 137 |
+
"509430392/train2": 0.0,
|
| 138 |
+
"585904130/train0": 0.0,
|
| 139 |
+
"585904130/train1": 0.0,
|
| 140 |
+
"585904130/train2": 0.0,
|
| 141 |
+
"608016290/train0": 0.0,
|
| 142 |
+
"608016290/train1": 0.0,
|
| 143 |
+
"608016290/train2": 0.0,
|
| 144 |
+
"626486813/train0": 0.0,
|
| 145 |
+
"626486813/train1": 0.0,
|
| 146 |
+
"626486813/train2": 0.0,
|
| 147 |
+
"636680137/train0": 0.0,
|
| 148 |
+
"636680137/train1": 0.0,
|
| 149 |
+
"636680137/train2": 0.0,
|
| 150 |
+
"655884605/train0": 0.0,
|
| 151 |
+
"655884605/train1": 0.0,
|
| 152 |
+
"655884605/train2": 0.0,
|
| 153 |
+
"672328262/train0": 0.0,
|
| 154 |
+
"672328262/train1": 0.0,
|
| 155 |
+
"672328262/train2": 0.0,
|
| 156 |
+
"701667486/train0": 0.0,
|
| 157 |
+
"701667486/train1": 0.0,
|
| 158 |
+
"701667486/train2": 0.0,
|
| 159 |
+
"747260437/train0": 0.0,
|
| 160 |
+
"747260437/train1": 0.0,
|
| 161 |
+
"747260437/train2": 0.0,
|
| 162 |
+
"750448296/train0": 0.0,
|
| 163 |
+
"750448296/train1": 0.0,
|
| 164 |
+
"750448296/train2": 0.0,
|
| 165 |
+
"767747424/train0": 0.0,
|
| 166 |
+
"767747424/train1": 0.0,
|
| 167 |
+
"767747424/train2": 0.0
|
| 168 |
+
}
|
| 169 |
+
},
|
| 170 |
+
"macro_by_task": {
|
| 171 |
+
"ci": {
|
| 172 |
+
"confidence": 0.95,
|
| 173 |
+
"high": 0.0,
|
| 174 |
+
"low": 0.0,
|
| 175 |
+
"mean": 0.0,
|
| 176 |
+
"n": 5,
|
| 177 |
+
"num_samples": 1000
|
| 178 |
+
},
|
| 179 |
+
"mean": 0.0,
|
| 180 |
+
"per_task": {
|
| 181 |
+
"LiftPegUpright-v1": 0.0,
|
| 182 |
+
"PickCube-v1": 0.0,
|
| 183 |
+
"PullCube-v1": 0.0,
|
| 184 |
+
"PushCube-v1": 0.0,
|
| 185 |
+
"StackCube-v1": 0.0
|
| 186 |
+
}
|
| 187 |
+
},
|
| 188 |
+
"metric": "base_safety_label_known",
|
| 189 |
+
"micro": {
|
| 190 |
+
"confidence": 0.95,
|
| 191 |
+
"high": 0.0,
|
| 192 |
+
"low": 0.0,
|
| 193 |
+
"mean": 0.0,
|
| 194 |
+
"n": 144,
|
| 195 |
+
"num_samples": 1000
|
| 196 |
+
},
|
| 197 |
+
"n": 144
|
| 198 |
+
},
|
| 199 |
"base_success": {
|
| 200 |
"macro_by_seed": {
|
| 201 |
"ci": {
|
| 202 |
"confidence": 0.95,
|
| 203 |
+
"high": 0.375,
|
| 204 |
+
"low": 0.20625,
|
| 205 |
"mean": 0.2916666666666667,
|
| 206 |
"n": 120,
|
| 207 |
+
"num_samples": 1000
|
| 208 |
},
|
| 209 |
"mean": 0.2916666666666667,
|
| 210 |
"per_seed": {
|
|
|
|
| 333 |
"macro_by_task": {
|
| 334 |
"ci": {
|
| 335 |
"confidence": 0.95,
|
| 336 |
+
"high": 0.3003636363636364,
|
| 337 |
"low": 0.0808080808080808,
|
| 338 |
"mean": 0.1954747474747475,
|
| 339 |
"n": 5,
|
| 340 |
+
"num_samples": 1000
|
| 341 |
},
|
| 342 |
"mean": 0.1954747474747475,
|
| 343 |
"per_task": {
|
|
|
|
| 355 |
"low": 0.2152777777777778,
|
| 356 |
"mean": 0.2847222222222222,
|
| 357 |
"n": 144,
|
| 358 |
+
"num_samples": 1000
|
| 359 |
},
|
| 360 |
"n": 144
|
| 361 |
},
|
|
|
|
| 363 |
"macro_by_seed": {
|
| 364 |
"ci": {
|
| 365 |
"confidence": 0.95,
|
| 366 |
+
"high": 0.9354621305075448,
|
| 367 |
+
"low": 0.6380658436120333,
|
| 368 |
"mean": 0.7857063563108873,
|
| 369 |
"n": 120,
|
| 370 |
+
"num_samples": 1000
|
| 371 |
},
|
| 372 |
"mean": 0.7857063563108873,
|
| 373 |
"per_seed": {
|
|
|
|
| 496 |
"macro_by_task": {
|
| 497 |
"ci": {
|
| 498 |
"confidence": 0.95,
|
| 499 |
+
"high": 0.8461356421672935,
|
| 500 |
+
"low": 0.5744831916065887,
|
| 501 |
"mean": 0.7040124819563852,
|
| 502 |
"n": 5,
|
| 503 |
+
"num_samples": 1000
|
| 504 |
},
|
| 505 |
"mean": 0.7040124819563852,
|
| 506 |
"per_task": {
|
|
|
|
| 514 |
"metric": "base_utility",
|
| 515 |
"micro": {
|
| 516 |
"confidence": 0.95,
|
| 517 |
+
"high": 0.878822695341013,
|
| 518 |
+
"low": 0.6172765076335054,
|
| 519 |
"mean": 0.7477176273686281,
|
| 520 |
"n": 144,
|
| 521 |
+
"num_samples": 1000
|
| 522 |
},
|
| 523 |
"n": 144
|
| 524 |
},
|
|
|
|
| 526 |
"macro_by_seed": {
|
| 527 |
"ci": {
|
| 528 |
"confidence": 0.95,
|
| 529 |
+
"high": 0.7439352242830662,
|
| 530 |
+
"low": 0.4503282577239184,
|
| 531 |
"mean": 0.5889192757033419,
|
| 532 |
"n": 120,
|
| 533 |
+
"num_samples": 1000
|
| 534 |
},
|
| 535 |
"mean": 0.5889192757033419,
|
| 536 |
"per_seed": {
|
|
|
|
| 659 |
"macro_by_task": {
|
| 660 |
"ci": {
|
| 661 |
"confidence": 0.95,
|
| 662 |
+
"high": 0.9107744794793879,
|
| 663 |
+
"low": 0.30645303343608976,
|
| 664 |
"mean": 0.5837108520707774,
|
| 665 |
"n": 5,
|
| 666 |
+
"num_samples": 1000
|
| 667 |
},
|
| 668 |
"mean": 0.5837108520707774,
|
| 669 |
"per_task": {
|
|
|
|
| 677 |
"metric": "branch_car_at_8",
|
| 678 |
"micro": {
|
| 679 |
"confidence": 0.95,
|
| 680 |
+
"high": 0.706521274279415,
|
| 681 |
+
"low": 0.4344568868643061,
|
| 682 |
"mean": 0.5604771331579994,
|
| 683 |
"n": 144,
|
| 684 |
+
"num_samples": 1000
|
| 685 |
+
},
|
| 686 |
+
"n": 144
|
| 687 |
+
},
|
| 688 |
+
"generated_safety_label_coverage_at_8": {
|
| 689 |
+
"macro_by_seed": {
|
| 690 |
+
"ci": {
|
| 691 |
+
"confidence": 0.95,
|
| 692 |
+
"high": 0.0,
|
| 693 |
+
"low": 0.0,
|
| 694 |
+
"mean": 0.0,
|
| 695 |
+
"n": 120,
|
| 696 |
+
"num_samples": 1000
|
| 697 |
+
},
|
| 698 |
+
"mean": 0.0,
|
| 699 |
+
"per_seed": {
|
| 700 |
+
"1073161778/train0": 0.0,
|
| 701 |
+
"1073161778/train1": 0.0,
|
| 702 |
+
"1073161778/train2": 0.0,
|
| 703 |
+
"1113371261/train0": 0.0,
|
| 704 |
+
"1113371261/train1": 0.0,
|
| 705 |
+
"1113371261/train2": 0.0,
|
| 706 |
+
"1151410346/train0": 0.0,
|
| 707 |
+
"1151410346/train1": 0.0,
|
| 708 |
+
"1151410346/train2": 0.0,
|
| 709 |
+
"1261930456/train0": 0.0,
|
| 710 |
+
"1261930456/train1": 0.0,
|
| 711 |
+
"1261930456/train2": 0.0,
|
| 712 |
+
"1324133122/train0": 0.0,
|
| 713 |
+
"1324133122/train1": 0.0,
|
| 714 |
+
"1324133122/train2": 0.0,
|
| 715 |
+
"1324642411/train0": 0.0,
|
| 716 |
+
"1324642411/train1": 0.0,
|
| 717 |
+
"1324642411/train2": 0.0,
|
| 718 |
+
"1375492786/train0": 0.0,
|
| 719 |
+
"1375492786/train1": 0.0,
|
| 720 |
+
"1375492786/train2": 0.0,
|
| 721 |
+
"141984615/train0": 0.0,
|
| 722 |
+
"141984615/train1": 0.0,
|
| 723 |
+
"141984615/train2": 0.0,
|
| 724 |
+
"1461069713/train0": 0.0,
|
| 725 |
+
"1461069713/train1": 0.0,
|
| 726 |
+
"1461069713/train2": 0.0,
|
| 727 |
+
"1480856810/train0": 0.0,
|
| 728 |
+
"1480856810/train1": 0.0,
|
| 729 |
+
"1480856810/train2": 0.0,
|
| 730 |
+
"1509980664/train0": 0.0,
|
| 731 |
+
"1509980664/train1": 0.0,
|
| 732 |
+
"1509980664/train2": 0.0,
|
| 733 |
+
"1522272686/train0": 0.0,
|
| 734 |
+
"1522272686/train1": 0.0,
|
| 735 |
+
"1522272686/train2": 0.0,
|
| 736 |
+
"1549324805/train0": 0.0,
|
| 737 |
+
"1549324805/train1": 0.0,
|
| 738 |
+
"1549324805/train2": 0.0,
|
| 739 |
+
"1578990937/train0": 0.0,
|
| 740 |
+
"1578990937/train1": 0.0,
|
| 741 |
+
"1578990937/train2": 0.0,
|
| 742 |
+
"1747178030/train0": 0.0,
|
| 743 |
+
"1747178030/train1": 0.0,
|
| 744 |
+
"1747178030/train2": 0.0,
|
| 745 |
+
"1790188787/train0": 0.0,
|
| 746 |
+
"1790188787/train1": 0.0,
|
| 747 |
+
"1790188787/train2": 0.0,
|
| 748 |
+
"1822210471/train0": 0.0,
|
| 749 |
+
"1822210471/train1": 0.0,
|
| 750 |
+
"1822210471/train2": 0.0,
|
| 751 |
+
"1845489013/train0": 0.0,
|
| 752 |
+
"1845489013/train1": 0.0,
|
| 753 |
+
"1845489013/train2": 0.0,
|
| 754 |
+
"1999022574/train0": 0.0,
|
| 755 |
+
"1999022574/train1": 0.0,
|
| 756 |
+
"1999022574/train2": 0.0,
|
| 757 |
+
"2012211287/train0": 0.0,
|
| 758 |
+
"2012211287/train1": 0.0,
|
| 759 |
+
"2012211287/train2": 0.0,
|
| 760 |
+
"2038594191/train0": 0.0,
|
| 761 |
+
"2038594191/train1": 0.0,
|
| 762 |
+
"2038594191/train2": 0.0,
|
| 763 |
+
"2054899126/train0": 0.0,
|
| 764 |
+
"2054899126/train1": 0.0,
|
| 765 |
+
"2054899126/train2": 0.0,
|
| 766 |
+
"2061635888/train0": 0.0,
|
| 767 |
+
"2061635888/train1": 0.0,
|
| 768 |
+
"2061635888/train2": 0.0,
|
| 769 |
+
"211245842/train0": 0.0,
|
| 770 |
+
"211245842/train1": 0.0,
|
| 771 |
+
"211245842/train2": 0.0,
|
| 772 |
+
"278008524/train0": 0.0,
|
| 773 |
+
"278008524/train1": 0.0,
|
| 774 |
+
"278008524/train2": 0.0,
|
| 775 |
+
"307068879/train0": 0.0,
|
| 776 |
+
"307068879/train1": 0.0,
|
| 777 |
+
"307068879/train2": 0.0,
|
| 778 |
+
"412102563/train0": 0.0,
|
| 779 |
+
"412102563/train1": 0.0,
|
| 780 |
+
"412102563/train2": 0.0,
|
| 781 |
+
"440577711/train0": 0.0,
|
| 782 |
+
"440577711/train1": 0.0,
|
| 783 |
+
"440577711/train2": 0.0,
|
| 784 |
+
"441046872/train0": 0.0,
|
| 785 |
+
"441046872/train1": 0.0,
|
| 786 |
+
"441046872/train2": 0.0,
|
| 787 |
+
"509430392/train0": 0.0,
|
| 788 |
+
"509430392/train1": 0.0,
|
| 789 |
+
"509430392/train2": 0.0,
|
| 790 |
+
"585904130/train0": 0.0,
|
| 791 |
+
"585904130/train1": 0.0,
|
| 792 |
+
"585904130/train2": 0.0,
|
| 793 |
+
"608016290/train0": 0.0,
|
| 794 |
+
"608016290/train1": 0.0,
|
| 795 |
+
"608016290/train2": 0.0,
|
| 796 |
+
"626486813/train0": 0.0,
|
| 797 |
+
"626486813/train1": 0.0,
|
| 798 |
+
"626486813/train2": 0.0,
|
| 799 |
+
"636680137/train0": 0.0,
|
| 800 |
+
"636680137/train1": 0.0,
|
| 801 |
+
"636680137/train2": 0.0,
|
| 802 |
+
"655884605/train0": 0.0,
|
| 803 |
+
"655884605/train1": 0.0,
|
| 804 |
+
"655884605/train2": 0.0,
|
| 805 |
+
"672328262/train0": 0.0,
|
| 806 |
+
"672328262/train1": 0.0,
|
| 807 |
+
"672328262/train2": 0.0,
|
| 808 |
+
"701667486/train0": 0.0,
|
| 809 |
+
"701667486/train1": 0.0,
|
| 810 |
+
"701667486/train2": 0.0,
|
| 811 |
+
"747260437/train0": 0.0,
|
| 812 |
+
"747260437/train1": 0.0,
|
| 813 |
+
"747260437/train2": 0.0,
|
| 814 |
+
"750448296/train0": 0.0,
|
| 815 |
+
"750448296/train1": 0.0,
|
| 816 |
+
"750448296/train2": 0.0,
|
| 817 |
+
"767747424/train0": 0.0,
|
| 818 |
+
"767747424/train1": 0.0,
|
| 819 |
+
"767747424/train2": 0.0
|
| 820 |
+
}
|
| 821 |
+
},
|
| 822 |
+
"macro_by_task": {
|
| 823 |
+
"ci": {
|
| 824 |
+
"confidence": 0.95,
|
| 825 |
+
"high": 0.0,
|
| 826 |
+
"low": 0.0,
|
| 827 |
+
"mean": 0.0,
|
| 828 |
+
"n": 5,
|
| 829 |
+
"num_samples": 1000
|
| 830 |
+
},
|
| 831 |
+
"mean": 0.0,
|
| 832 |
+
"per_task": {
|
| 833 |
+
"LiftPegUpright-v1": 0.0,
|
| 834 |
+
"PickCube-v1": 0.0,
|
| 835 |
+
"PullCube-v1": 0.0,
|
| 836 |
+
"PushCube-v1": 0.0,
|
| 837 |
+
"StackCube-v1": 0.0
|
| 838 |
+
}
|
| 839 |
+
},
|
| 840 |
+
"metric": "generated_safety_label_coverage_at_8",
|
| 841 |
+
"micro": {
|
| 842 |
+
"confidence": 0.95,
|
| 843 |
+
"high": 0.0,
|
| 844 |
+
"low": 0.0,
|
| 845 |
+
"mean": 0.0,
|
| 846 |
+
"n": 144,
|
| 847 |
+
"num_samples": 1000
|
| 848 |
},
|
| 849 |
"n": 144
|
| 850 |
},
|
|
|
|
| 853 |
"ci": {
|
| 854 |
"confidence": 0.95,
|
| 855 |
"high": 0.8555555555555556,
|
| 856 |
+
"low": 0.7180555555555556,
|
| 857 |
"mean": 0.7854166666666668,
|
| 858 |
"n": 120,
|
| 859 |
+
"num_samples": 1000
|
| 860 |
},
|
| 861 |
"mean": 0.7854166666666668,
|
| 862 |
"per_seed": {
|
|
|
|
| 989 |
"low": 0.6786666666666668,
|
| 990 |
"mean": 0.8706666666666667,
|
| 991 |
"n": 5,
|
| 992 |
+
"num_samples": 1000
|
| 993 |
},
|
| 994 |
"mean": 0.8706666666666667,
|
| 995 |
"per_task": {
|
|
|
|
| 1003 |
"metric": "hidden_chart_oracle_success_at_8",
|
| 1004 |
"micro": {
|
| 1005 |
"confidence": 0.95,
|
| 1006 |
+
"high": 0.7986111111111112,
|
| 1007 |
"low": 0.6597222222222222,
|
| 1008 |
"mean": 0.7291666666666666,
|
| 1009 |
"n": 144,
|
| 1010 |
+
"num_samples": 1000
|
| 1011 |
},
|
| 1012 |
"n": 144
|
| 1013 |
},
|
|
|
|
| 1015 |
"macro_by_seed": {
|
| 1016 |
"ci": {
|
| 1017 |
"confidence": 0.95,
|
| 1018 |
+
"high": 1.7441341616213322,
|
| 1019 |
+
"low": 1.49932806055165,
|
| 1020 |
"mean": 1.6190625454609595,
|
| 1021 |
"n": 120,
|
| 1022 |
+
"num_samples": 1000
|
| 1023 |
},
|
| 1024 |
"mean": 1.6190625454609595,
|
| 1025 |
"per_seed": {
|
|
|
|
| 1152 |
"low": 1.428388743619124,
|
| 1153 |
"mean": 1.7696735892097153,
|
| 1154 |
"n": 5,
|
| 1155 |
+
"num_samples": 1000
|
| 1156 |
},
|
| 1157 |
"mean": 1.7696735892097153,
|
| 1158 |
"per_task": {
|
|
|
|
| 1166 |
"metric": "hidden_chart_oracle_utility_at_8",
|
| 1167 |
"micro": {
|
| 1168 |
"confidence": 0.95,
|
| 1169 |
+
"high": 1.643800659933024,
|
| 1170 |
+
"low": 1.3941160800556343,
|
| 1171 |
"mean": 1.5183161981403828,
|
| 1172 |
"n": 144,
|
| 1173 |
+
"num_samples": 1000
|
| 1174 |
},
|
| 1175 |
"n": 144
|
| 1176 |
},
|
|
|
|
| 1178 |
"macro_by_seed": {
|
| 1179 |
"ci": {
|
| 1180 |
"confidence": 0.95,
|
| 1181 |
+
"high": 0.6125,
|
| 1182 |
"low": 0.43680555555555556,
|
| 1183 |
"mean": 0.5243055555555556,
|
| 1184 |
"n": 120,
|
| 1185 |
+
"num_samples": 1000
|
| 1186 |
},
|
| 1187 |
"mean": 0.5243055555555556,
|
| 1188 |
"per_seed": {
|
|
|
|
| 1311 |
"macro_by_task": {
|
| 1312 |
"ci": {
|
| 1313 |
"confidence": 0.95,
|
| 1314 |
+
"high": 0.7288888888888889,
|
| 1315 |
+
"low": 0.22399999999999998,
|
| 1316 |
"mean": 0.5128888888888888,
|
| 1317 |
"n": 5,
|
| 1318 |
+
"num_samples": 1000
|
| 1319 |
},
|
| 1320 |
"mean": 0.5128888888888888,
|
| 1321 |
"per_task": {
|
|
|
|
| 1330 |
"micro": {
|
| 1331 |
"confidence": 0.95,
|
| 1332 |
"high": 0.6041666666666666,
|
| 1333 |
+
"low": 0.4513888888888889,
|
| 1334 |
"mean": 0.5277777777777778,
|
| 1335 |
"n": 144,
|
| 1336 |
+
"num_samples": 1000
|
| 1337 |
},
|
| 1338 |
"n": 144
|
| 1339 |
},
|
|
|
|
| 1341 |
"macro_by_seed": {
|
| 1342 |
"ci": {
|
| 1343 |
"confidence": 0.95,
|
| 1344 |
+
"high": 0.19196862734608558,
|
| 1345 |
+
"low": 0.15260038495896353,
|
| 1346 |
"mean": 0.17180047070386753,
|
| 1347 |
"n": 120,
|
| 1348 |
+
"num_samples": 1000
|
| 1349 |
},
|
| 1350 |
"mean": 0.17180047070386753,
|
| 1351 |
"per_seed": {
|
|
|
|
| 1431 |
"412102563/train1": 0.05497825758025142,
|
| 1432 |
"412102563/train2": 0.23421550754433385,
|
| 1433 |
"440577711/train0": 0.0,
|
| 1434 |
+
"440577711/train1": 0.08649062167899924,
|
| 1435 |
+
"440577711/train2": 0.15795943889155128,
|
| 1436 |
+
"441046872/train0": 0.07559488024963246,
|
| 1437 |
+
"441046872/train1": 0.03055053741629537,
|
| 1438 |
+
"441046872/train2": 0.10573947390621682,
|
| 1439 |
+
"509430392/train0": 0.05599945856896821,
|
| 1440 |
+
"509430392/train1": 0.29083191353306787,
|
| 1441 |
+
"509430392/train2": 0.15055463357076268,
|
| 1442 |
+
"585904130/train0": 0.14940863096718615,
|
| 1443 |
+
"585904130/train1": 0.24258759953449954,
|
| 1444 |
+
"585904130/train2": 0.31468676305833865,
|
| 1445 |
+
"608016290/train0": 0.3387708317969876,
|
| 1446 |
+
"608016290/train1": 0.1955644997142938,
|
| 1447 |
+
"608016290/train2": 0.159510713855325,
|
| 1448 |
+
"626486813/train0": 0.21441221141580127,
|
| 1449 |
+
"626486813/train1": 0.170949273586752,
|
| 1450 |
+
"626486813/train2": 0.16604576952462607,
|
| 1451 |
+
"636680137/train0": 0.2814239671003126,
|
| 1452 |
+
"636680137/train1": 0.2625830101532915,
|
| 1453 |
+
"636680137/train2": 0.12020315308522647,
|
| 1454 |
+
"655884605/train0": 0.14628351854390587,
|
| 1455 |
+
"655884605/train1": 0.0690564909179267,
|
| 1456 |
+
"655884605/train2": 0.320973287648995,
|
| 1457 |
+
"672328262/train0": 0.16662194482040366,
|
| 1458 |
+
"672328262/train1": 0.18784609006259695,
|
| 1459 |
+
"672328262/train2": 0.2998508752953551,
|
| 1460 |
+
"701667486/train0": 0.13404567006107426,
|
| 1461 |
+
"701667486/train1": 0.12687721623280993,
|
| 1462 |
+
"701667486/train2": 0.0,
|
| 1463 |
+
"747260437/train0": 0.14673632710621276,
|
| 1464 |
+
"747260437/train1": 0.12246432985699784,
|
| 1465 |
+
"747260437/train2": 0.04243786608353012,
|
| 1466 |
+
"750448296/train0": 0.193706224527593,
|
| 1467 |
+
"750448296/train1": 0.3180339607786489,
|
| 1468 |
+
"750448296/train2": 0.20399296184071347,
|
| 1469 |
+
"767747424/train0": 0.20750561576560225,
|
| 1470 |
+
"767747424/train1": 0.14443725209498554,
|
| 1471 |
+
"767747424/train2": 0.1864480763603019
|
| 1472 |
+
}
|
| 1473 |
+
},
|
| 1474 |
+
"macro_by_task": {
|
| 1475 |
+
"ci": {
|
| 1476 |
+
"confidence": 0.95,
|
| 1477 |
+
"high": 0.20659742047874713,
|
| 1478 |
+
"low": 0.16666709804041427,
|
| 1479 |
+
"mean": 0.18195680108276988,
|
| 1480 |
+
"n": 5,
|
| 1481 |
+
"num_samples": 1000
|
| 1482 |
+
},
|
| 1483 |
+
"mean": 0.18195680108276988,
|
| 1484 |
+
"per_task": {
|
| 1485 |
+
"LiftPegUpright-v1": 0.1621114782545992,
|
| 1486 |
+
"PickCube-v1": 0.2290622239534128,
|
| 1487 |
+
"PullCube-v1": 0.17280987267234024,
|
| 1488 |
+
"PushCube-v1": 0.1716092477675637,
|
| 1489 |
+
"StackCube-v1": 0.1741911827659335
|
| 1490 |
+
}
|
| 1491 |
+
},
|
| 1492 |
+
"metric": "pairwise_causal_calibration_ece",
|
| 1493 |
+
"micro": {
|
| 1494 |
+
"confidence": 0.95,
|
| 1495 |
+
"high": 0.19105138497885166,
|
| 1496 |
+
"low": 0.15467712413428017,
|
| 1497 |
+
"mean": 0.173033852688866,
|
| 1498 |
+
"n": 144,
|
| 1499 |
+
"num_samples": 1000
|
| 1500 |
+
},
|
| 1501 |
+
"n": 144
|
| 1502 |
+
},
|
| 1503 |
+
"proposal_oracle_safety_label_known_at_8": {
|
| 1504 |
+
"macro_by_seed": {
|
| 1505 |
+
"ci": {
|
| 1506 |
+
"confidence": 0.95,
|
| 1507 |
+
"high": 0.0,
|
| 1508 |
+
"low": 0.0,
|
| 1509 |
+
"mean": 0.0,
|
| 1510 |
+
"n": 120,
|
| 1511 |
+
"num_samples": 1000
|
| 1512 |
+
},
|
| 1513 |
+
"mean": 0.0,
|
| 1514 |
+
"per_seed": {
|
| 1515 |
+
"1073161778/train0": 0.0,
|
| 1516 |
+
"1073161778/train1": 0.0,
|
| 1517 |
+
"1073161778/train2": 0.0,
|
| 1518 |
+
"1113371261/train0": 0.0,
|
| 1519 |
+
"1113371261/train1": 0.0,
|
| 1520 |
+
"1113371261/train2": 0.0,
|
| 1521 |
+
"1151410346/train0": 0.0,
|
| 1522 |
+
"1151410346/train1": 0.0,
|
| 1523 |
+
"1151410346/train2": 0.0,
|
| 1524 |
+
"1261930456/train0": 0.0,
|
| 1525 |
+
"1261930456/train1": 0.0,
|
| 1526 |
+
"1261930456/train2": 0.0,
|
| 1527 |
+
"1324133122/train0": 0.0,
|
| 1528 |
+
"1324133122/train1": 0.0,
|
| 1529 |
+
"1324133122/train2": 0.0,
|
| 1530 |
+
"1324642411/train0": 0.0,
|
| 1531 |
+
"1324642411/train1": 0.0,
|
| 1532 |
+
"1324642411/train2": 0.0,
|
| 1533 |
+
"1375492786/train0": 0.0,
|
| 1534 |
+
"1375492786/train1": 0.0,
|
| 1535 |
+
"1375492786/train2": 0.0,
|
| 1536 |
+
"141984615/train0": 0.0,
|
| 1537 |
+
"141984615/train1": 0.0,
|
| 1538 |
+
"141984615/train2": 0.0,
|
| 1539 |
+
"1461069713/train0": 0.0,
|
| 1540 |
+
"1461069713/train1": 0.0,
|
| 1541 |
+
"1461069713/train2": 0.0,
|
| 1542 |
+
"1480856810/train0": 0.0,
|
| 1543 |
+
"1480856810/train1": 0.0,
|
| 1544 |
+
"1480856810/train2": 0.0,
|
| 1545 |
+
"1509980664/train0": 0.0,
|
| 1546 |
+
"1509980664/train1": 0.0,
|
| 1547 |
+
"1509980664/train2": 0.0,
|
| 1548 |
+
"1522272686/train0": 0.0,
|
| 1549 |
+
"1522272686/train1": 0.0,
|
| 1550 |
+
"1522272686/train2": 0.0,
|
| 1551 |
+
"1549324805/train0": 0.0,
|
| 1552 |
+
"1549324805/train1": 0.0,
|
| 1553 |
+
"1549324805/train2": 0.0,
|
| 1554 |
+
"1578990937/train0": 0.0,
|
| 1555 |
+
"1578990937/train1": 0.0,
|
| 1556 |
+
"1578990937/train2": 0.0,
|
| 1557 |
+
"1747178030/train0": 0.0,
|
| 1558 |
+
"1747178030/train1": 0.0,
|
| 1559 |
+
"1747178030/train2": 0.0,
|
| 1560 |
+
"1790188787/train0": 0.0,
|
| 1561 |
+
"1790188787/train1": 0.0,
|
| 1562 |
+
"1790188787/train2": 0.0,
|
| 1563 |
+
"1822210471/train0": 0.0,
|
| 1564 |
+
"1822210471/train1": 0.0,
|
| 1565 |
+
"1822210471/train2": 0.0,
|
| 1566 |
+
"1845489013/train0": 0.0,
|
| 1567 |
+
"1845489013/train1": 0.0,
|
| 1568 |
+
"1845489013/train2": 0.0,
|
| 1569 |
+
"1999022574/train0": 0.0,
|
| 1570 |
+
"1999022574/train1": 0.0,
|
| 1571 |
+
"1999022574/train2": 0.0,
|
| 1572 |
+
"2012211287/train0": 0.0,
|
| 1573 |
+
"2012211287/train1": 0.0,
|
| 1574 |
+
"2012211287/train2": 0.0,
|
| 1575 |
+
"2038594191/train0": 0.0,
|
| 1576 |
+
"2038594191/train1": 0.0,
|
| 1577 |
+
"2038594191/train2": 0.0,
|
| 1578 |
+
"2054899126/train0": 0.0,
|
| 1579 |
+
"2054899126/train1": 0.0,
|
| 1580 |
+
"2054899126/train2": 0.0,
|
| 1581 |
+
"2061635888/train0": 0.0,
|
| 1582 |
+
"2061635888/train1": 0.0,
|
| 1583 |
+
"2061635888/train2": 0.0,
|
| 1584 |
+
"211245842/train0": 0.0,
|
| 1585 |
+
"211245842/train1": 0.0,
|
| 1586 |
+
"211245842/train2": 0.0,
|
| 1587 |
+
"278008524/train0": 0.0,
|
| 1588 |
+
"278008524/train1": 0.0,
|
| 1589 |
+
"278008524/train2": 0.0,
|
| 1590 |
+
"307068879/train0": 0.0,
|
| 1591 |
+
"307068879/train1": 0.0,
|
| 1592 |
+
"307068879/train2": 0.0,
|
| 1593 |
+
"412102563/train0": 0.0,
|
| 1594 |
+
"412102563/train1": 0.0,
|
| 1595 |
+
"412102563/train2": 0.0,
|
| 1596 |
+
"440577711/train0": 0.0,
|
| 1597 |
+
"440577711/train1": 0.0,
|
| 1598 |
+
"440577711/train2": 0.0,
|
| 1599 |
+
"441046872/train0": 0.0,
|
| 1600 |
+
"441046872/train1": 0.0,
|
| 1601 |
+
"441046872/train2": 0.0,
|
| 1602 |
+
"509430392/train0": 0.0,
|
| 1603 |
+
"509430392/train1": 0.0,
|
| 1604 |
+
"509430392/train2": 0.0,
|
| 1605 |
+
"585904130/train0": 0.0,
|
| 1606 |
+
"585904130/train1": 0.0,
|
| 1607 |
+
"585904130/train2": 0.0,
|
| 1608 |
+
"608016290/train0": 0.0,
|
| 1609 |
+
"608016290/train1": 0.0,
|
| 1610 |
+
"608016290/train2": 0.0,
|
| 1611 |
+
"626486813/train0": 0.0,
|
| 1612 |
+
"626486813/train1": 0.0,
|
| 1613 |
+
"626486813/train2": 0.0,
|
| 1614 |
+
"636680137/train0": 0.0,
|
| 1615 |
+
"636680137/train1": 0.0,
|
| 1616 |
+
"636680137/train2": 0.0,
|
| 1617 |
+
"655884605/train0": 0.0,
|
| 1618 |
+
"655884605/train1": 0.0,
|
| 1619 |
+
"655884605/train2": 0.0,
|
| 1620 |
+
"672328262/train0": 0.0,
|
| 1621 |
+
"672328262/train1": 0.0,
|
| 1622 |
+
"672328262/train2": 0.0,
|
| 1623 |
+
"701667486/train0": 0.0,
|
| 1624 |
+
"701667486/train1": 0.0,
|
| 1625 |
"701667486/train2": 0.0,
|
| 1626 |
+
"747260437/train0": 0.0,
|
| 1627 |
+
"747260437/train1": 0.0,
|
| 1628 |
+
"747260437/train2": 0.0,
|
| 1629 |
+
"750448296/train0": 0.0,
|
| 1630 |
+
"750448296/train1": 0.0,
|
| 1631 |
+
"750448296/train2": 0.0,
|
| 1632 |
+
"767747424/train0": 0.0,
|
| 1633 |
+
"767747424/train1": 0.0,
|
| 1634 |
+
"767747424/train2": 0.0
|
| 1635 |
}
|
| 1636 |
},
|
| 1637 |
"macro_by_task": {
|
| 1638 |
"ci": {
|
| 1639 |
"confidence": 0.95,
|
| 1640 |
+
"high": 0.0,
|
| 1641 |
+
"low": 0.0,
|
| 1642 |
+
"mean": 0.0,
|
| 1643 |
"n": 5,
|
| 1644 |
+
"num_samples": 1000
|
| 1645 |
},
|
| 1646 |
+
"mean": 0.0,
|
| 1647 |
"per_task": {
|
| 1648 |
+
"LiftPegUpright-v1": 0.0,
|
| 1649 |
+
"PickCube-v1": 0.0,
|
| 1650 |
+
"PullCube-v1": 0.0,
|
| 1651 |
+
"PushCube-v1": 0.0,
|
| 1652 |
+
"StackCube-v1": 0.0
|
| 1653 |
}
|
| 1654 |
},
|
| 1655 |
+
"metric": "proposal_oracle_safety_label_known_at_8",
|
| 1656 |
"micro": {
|
| 1657 |
"confidence": 0.95,
|
| 1658 |
+
"high": 0.0,
|
| 1659 |
+
"low": 0.0,
|
| 1660 |
+
"mean": 0.0,
|
| 1661 |
"n": 144,
|
| 1662 |
+
"num_samples": 1000
|
| 1663 |
},
|
| 1664 |
"n": 144
|
| 1665 |
},
|
|
|
|
| 1667 |
"macro_by_seed": {
|
| 1668 |
"ci": {
|
| 1669 |
"confidence": 0.95,
|
| 1670 |
+
"high": 0.6194444444444444,
|
| 1671 |
+
"low": 0.4444444444444445,
|
| 1672 |
"mean": 0.53125,
|
| 1673 |
"n": 120,
|
| 1674 |
+
"num_samples": 1000
|
| 1675 |
},
|
| 1676 |
"mean": 0.53125,
|
| 1677 |
"per_seed": {
|
|
|
|
| 1800 |
"macro_by_task": {
|
| 1801 |
"ci": {
|
| 1802 |
"confidence": 0.95,
|
| 1803 |
+
"high": 0.6906666666666667,
|
| 1804 |
"low": 0.21818181818181817,
|
| 1805 |
"mean": 0.4755151515151515,
|
| 1806 |
"n": 5,
|
| 1807 |
+
"num_samples": 1000
|
| 1808 |
},
|
| 1809 |
"mean": 0.4755151515151515,
|
| 1810 |
"per_task": {
|
|
|
|
| 1818 |
"metric": "proposal_oracle_success_at_8",
|
| 1819 |
"micro": {
|
| 1820 |
"confidence": 0.95,
|
| 1821 |
+
"high": 0.5902777777777778,
|
| 1822 |
"low": 0.4305555555555556,
|
| 1823 |
"mean": 0.5069444444444444,
|
| 1824 |
"n": 144,
|
| 1825 |
+
"num_samples": 1000
|
| 1826 |
},
|
| 1827 |
"n": 144
|
| 1828 |
},
|
|
|
|
| 1830 |
"macro_by_seed": {
|
| 1831 |
"ci": {
|
| 1832 |
"confidence": 0.95,
|
| 1833 |
+
"high": 0.33888888888888885,
|
| 1834 |
"low": 0.14513888888888887,
|
| 1835 |
"mean": 0.23958333333333334,
|
| 1836 |
"n": 120,
|
| 1837 |
+
"num_samples": 1000
|
| 1838 |
},
|
| 1839 |
"mean": 0.23958333333333334,
|
| 1840 |
"per_seed": {
|
|
|
|
| 1964 |
"ci": {
|
| 1965 |
"confidence": 0.95,
|
| 1966 |
"high": 0.4760000000000001,
|
| 1967 |
+
"low": 0.09333333333333334,
|
| 1968 |
"mean": 0.28004040404040403,
|
| 1969 |
"n": 5,
|
| 1970 |
+
"num_samples": 1000
|
| 1971 |
},
|
| 1972 |
"mean": 0.28004040404040403,
|
| 1973 |
"per_task": {
|
|
|
|
| 1981 |
"metric": "proposal_oracle_success_gain_over_base_at_8",
|
| 1982 |
"micro": {
|
| 1983 |
"confidence": 0.95,
|
| 1984 |
+
"high": 0.3055555555555556,
|
| 1985 |
"low": 0.13194444444444445,
|
| 1986 |
"mean": 0.2222222222222222,
|
| 1987 |
"n": 144,
|
| 1988 |
+
"num_samples": 1000
|
| 1989 |
},
|
| 1990 |
"n": 144
|
| 1991 |
},
|
|
|
|
| 1993 |
"macro_by_seed": {
|
| 1994 |
"ci": {
|
| 1995 |
"confidence": 0.95,
|
| 1996 |
+
"high": 1.3228005942080117,
|
| 1997 |
+
"low": 1.0087994246174476,
|
| 1998 |
"mean": 1.1654707208521562,
|
| 1999 |
"n": 120,
|
| 2000 |
+
"num_samples": 1000
|
| 2001 |
},
|
| 2002 |
"mean": 1.1654707208521562,
|
| 2003 |
"per_seed": {
|
|
|
|
| 2126 |
"macro_by_task": {
|
| 2127 |
"ci": {
|
| 2128 |
"confidence": 0.95,
|
| 2129 |
+
"high": 1.4793404242864086,
|
| 2130 |
"low": 0.6483407714106143,
|
| 2131 |
"mean": 1.0926857944947828,
|
| 2132 |
"n": 5,
|
| 2133 |
+
"num_samples": 1000
|
| 2134 |
},
|
| 2135 |
"mean": 1.0926857944947828,
|
| 2136 |
"per_task": {
|
|
|
|
| 2144 |
"metric": "proposal_oracle_utility_at_8",
|
| 2145 |
"micro": {
|
| 2146 |
"confidence": 0.95,
|
| 2147 |
+
"high": 1.266212323441222,
|
| 2148 |
+
"low": 0.9685431999128519,
|
| 2149 |
"mean": 1.109000739677059,
|
| 2150 |
"n": 144,
|
| 2151 |
+
"num_samples": 1000
|
| 2152 |
},
|
| 2153 |
"n": 144
|
| 2154 |
},
|
|
|
|
| 2156 |
"macro_by_seed": {
|
| 2157 |
"ci": {
|
| 2158 |
"confidence": 0.95,
|
| 2159 |
+
"high": 0.5465957584145751,
|
| 2160 |
+
"low": 0.21765759116824485,
|
| 2161 |
"mean": 0.37976436454126894,
|
| 2162 |
"n": 120,
|
| 2163 |
+
"num_samples": 1000
|
| 2164 |
},
|
| 2165 |
"mean": 0.37976436454126894,
|
| 2166 |
"per_seed": {
|
|
|
|
| 2289 |
"macro_by_task": {
|
| 2290 |
"ci": {
|
| 2291 |
"confidence": 0.95,
|
| 2292 |
+
"high": 0.8288499944077596,
|
| 2293 |
+
"low": -0.05187151841664066,
|
| 2294 |
"mean": 0.3886733125383975,
|
| 2295 |
"n": 5,
|
| 2296 |
+
"num_samples": 1000
|
| 2297 |
},
|
| 2298 |
"mean": 0.3886733125383975,
|
| 2299 |
"per_task": {
|
|
|
|
| 2307 |
"metric": "proposal_oracle_utility_gain_over_base_at_8",
|
| 2308 |
"micro": {
|
| 2309 |
"confidence": 0.95,
|
| 2310 |
+
"high": 0.5091405405780987,
|
| 2311 |
+
"low": 0.2108570472773863,
|
| 2312 |
"mean": 0.36128311230843085,
|
| 2313 |
"n": 144,
|
| 2314 |
+
"num_samples": 1000
|
| 2315 |
+
},
|
| 2316 |
+
"n": 144
|
| 2317 |
+
},
|
| 2318 |
+
"selected_safety_label_known_at_8": {
|
| 2319 |
+
"macro_by_seed": {
|
| 2320 |
+
"ci": {
|
| 2321 |
+
"confidence": 0.95,
|
| 2322 |
+
"high": 0.0,
|
| 2323 |
+
"low": 0.0,
|
| 2324 |
+
"mean": 0.0,
|
| 2325 |
+
"n": 120,
|
| 2326 |
+
"num_samples": 1000
|
| 2327 |
+
},
|
| 2328 |
+
"mean": 0.0,
|
| 2329 |
+
"per_seed": {
|
| 2330 |
+
"1073161778/train0": 0.0,
|
| 2331 |
+
"1073161778/train1": 0.0,
|
| 2332 |
+
"1073161778/train2": 0.0,
|
| 2333 |
+
"1113371261/train0": 0.0,
|
| 2334 |
+
"1113371261/train1": 0.0,
|
| 2335 |
+
"1113371261/train2": 0.0,
|
| 2336 |
+
"1151410346/train0": 0.0,
|
| 2337 |
+
"1151410346/train1": 0.0,
|
| 2338 |
+
"1151410346/train2": 0.0,
|
| 2339 |
+
"1261930456/train0": 0.0,
|
| 2340 |
+
"1261930456/train1": 0.0,
|
| 2341 |
+
"1261930456/train2": 0.0,
|
| 2342 |
+
"1324133122/train0": 0.0,
|
| 2343 |
+
"1324133122/train1": 0.0,
|
| 2344 |
+
"1324133122/train2": 0.0,
|
| 2345 |
+
"1324642411/train0": 0.0,
|
| 2346 |
+
"1324642411/train1": 0.0,
|
| 2347 |
+
"1324642411/train2": 0.0,
|
| 2348 |
+
"1375492786/train0": 0.0,
|
| 2349 |
+
"1375492786/train1": 0.0,
|
| 2350 |
+
"1375492786/train2": 0.0,
|
| 2351 |
+
"141984615/train0": 0.0,
|
| 2352 |
+
"141984615/train1": 0.0,
|
| 2353 |
+
"141984615/train2": 0.0,
|
| 2354 |
+
"1461069713/train0": 0.0,
|
| 2355 |
+
"1461069713/train1": 0.0,
|
| 2356 |
+
"1461069713/train2": 0.0,
|
| 2357 |
+
"1480856810/train0": 0.0,
|
| 2358 |
+
"1480856810/train1": 0.0,
|
| 2359 |
+
"1480856810/train2": 0.0,
|
| 2360 |
+
"1509980664/train0": 0.0,
|
| 2361 |
+
"1509980664/train1": 0.0,
|
| 2362 |
+
"1509980664/train2": 0.0,
|
| 2363 |
+
"1522272686/train0": 0.0,
|
| 2364 |
+
"1522272686/train1": 0.0,
|
| 2365 |
+
"1522272686/train2": 0.0,
|
| 2366 |
+
"1549324805/train0": 0.0,
|
| 2367 |
+
"1549324805/train1": 0.0,
|
| 2368 |
+
"1549324805/train2": 0.0,
|
| 2369 |
+
"1578990937/train0": 0.0,
|
| 2370 |
+
"1578990937/train1": 0.0,
|
| 2371 |
+
"1578990937/train2": 0.0,
|
| 2372 |
+
"1747178030/train0": 0.0,
|
| 2373 |
+
"1747178030/train1": 0.0,
|
| 2374 |
+
"1747178030/train2": 0.0,
|
| 2375 |
+
"1790188787/train0": 0.0,
|
| 2376 |
+
"1790188787/train1": 0.0,
|
| 2377 |
+
"1790188787/train2": 0.0,
|
| 2378 |
+
"1822210471/train0": 0.0,
|
| 2379 |
+
"1822210471/train1": 0.0,
|
| 2380 |
+
"1822210471/train2": 0.0,
|
| 2381 |
+
"1845489013/train0": 0.0,
|
| 2382 |
+
"1845489013/train1": 0.0,
|
| 2383 |
+
"1845489013/train2": 0.0,
|
| 2384 |
+
"1999022574/train0": 0.0,
|
| 2385 |
+
"1999022574/train1": 0.0,
|
| 2386 |
+
"1999022574/train2": 0.0,
|
| 2387 |
+
"2012211287/train0": 0.0,
|
| 2388 |
+
"2012211287/train1": 0.0,
|
| 2389 |
+
"2012211287/train2": 0.0,
|
| 2390 |
+
"2038594191/train0": 0.0,
|
| 2391 |
+
"2038594191/train1": 0.0,
|
| 2392 |
+
"2038594191/train2": 0.0,
|
| 2393 |
+
"2054899126/train0": 0.0,
|
| 2394 |
+
"2054899126/train1": 0.0,
|
| 2395 |
+
"2054899126/train2": 0.0,
|
| 2396 |
+
"2061635888/train0": 0.0,
|
| 2397 |
+
"2061635888/train1": 0.0,
|
| 2398 |
+
"2061635888/train2": 0.0,
|
| 2399 |
+
"211245842/train0": 0.0,
|
| 2400 |
+
"211245842/train1": 0.0,
|
| 2401 |
+
"211245842/train2": 0.0,
|
| 2402 |
+
"278008524/train0": 0.0,
|
| 2403 |
+
"278008524/train1": 0.0,
|
| 2404 |
+
"278008524/train2": 0.0,
|
| 2405 |
+
"307068879/train0": 0.0,
|
| 2406 |
+
"307068879/train1": 0.0,
|
| 2407 |
+
"307068879/train2": 0.0,
|
| 2408 |
+
"412102563/train0": 0.0,
|
| 2409 |
+
"412102563/train1": 0.0,
|
| 2410 |
+
"412102563/train2": 0.0,
|
| 2411 |
+
"440577711/train0": 0.0,
|
| 2412 |
+
"440577711/train1": 0.0,
|
| 2413 |
+
"440577711/train2": 0.0,
|
| 2414 |
+
"441046872/train0": 0.0,
|
| 2415 |
+
"441046872/train1": 0.0,
|
| 2416 |
+
"441046872/train2": 0.0,
|
| 2417 |
+
"509430392/train0": 0.0,
|
| 2418 |
+
"509430392/train1": 0.0,
|
| 2419 |
+
"509430392/train2": 0.0,
|
| 2420 |
+
"585904130/train0": 0.0,
|
| 2421 |
+
"585904130/train1": 0.0,
|
| 2422 |
+
"585904130/train2": 0.0,
|
| 2423 |
+
"608016290/train0": 0.0,
|
| 2424 |
+
"608016290/train1": 0.0,
|
| 2425 |
+
"608016290/train2": 0.0,
|
| 2426 |
+
"626486813/train0": 0.0,
|
| 2427 |
+
"626486813/train1": 0.0,
|
| 2428 |
+
"626486813/train2": 0.0,
|
| 2429 |
+
"636680137/train0": 0.0,
|
| 2430 |
+
"636680137/train1": 0.0,
|
| 2431 |
+
"636680137/train2": 0.0,
|
| 2432 |
+
"655884605/train0": 0.0,
|
| 2433 |
+
"655884605/train1": 0.0,
|
| 2434 |
+
"655884605/train2": 0.0,
|
| 2435 |
+
"672328262/train0": 0.0,
|
| 2436 |
+
"672328262/train1": 0.0,
|
| 2437 |
+
"672328262/train2": 0.0,
|
| 2438 |
+
"701667486/train0": 0.0,
|
| 2439 |
+
"701667486/train1": 0.0,
|
| 2440 |
+
"701667486/train2": 0.0,
|
| 2441 |
+
"747260437/train0": 0.0,
|
| 2442 |
+
"747260437/train1": 0.0,
|
| 2443 |
+
"747260437/train2": 0.0,
|
| 2444 |
+
"750448296/train0": 0.0,
|
| 2445 |
+
"750448296/train1": 0.0,
|
| 2446 |
+
"750448296/train2": 0.0,
|
| 2447 |
+
"767747424/train0": 0.0,
|
| 2448 |
+
"767747424/train1": 0.0,
|
| 2449 |
+
"767747424/train2": 0.0
|
| 2450 |
+
}
|
| 2451 |
+
},
|
| 2452 |
+
"macro_by_task": {
|
| 2453 |
+
"ci": {
|
| 2454 |
+
"confidence": 0.95,
|
| 2455 |
+
"high": 0.0,
|
| 2456 |
+
"low": 0.0,
|
| 2457 |
+
"mean": 0.0,
|
| 2458 |
+
"n": 5,
|
| 2459 |
+
"num_samples": 1000
|
| 2460 |
+
},
|
| 2461 |
+
"mean": 0.0,
|
| 2462 |
+
"per_task": {
|
| 2463 |
+
"LiftPegUpright-v1": 0.0,
|
| 2464 |
+
"PickCube-v1": 0.0,
|
| 2465 |
+
"PullCube-v1": 0.0,
|
| 2466 |
+
"PushCube-v1": 0.0,
|
| 2467 |
+
"StackCube-v1": 0.0
|
| 2468 |
+
}
|
| 2469 |
+
},
|
| 2470 |
+
"metric": "selected_safety_label_known_at_8",
|
| 2471 |
+
"micro": {
|
| 2472 |
+
"confidence": 0.95,
|
| 2473 |
+
"high": 0.0,
|
| 2474 |
+
"low": 0.0,
|
| 2475 |
+
"mean": 0.0,
|
| 2476 |
+
"n": 144,
|
| 2477 |
+
"num_samples": 1000
|
| 2478 |
},
|
| 2479 |
"n": 144
|
| 2480 |
},
|
|
|
|
| 2483 |
"ci": {
|
| 2484 |
"confidence": 0.95,
|
| 2485 |
"high": 0.3055555555555555,
|
| 2486 |
+
"low": 0.15833333333333333,
|
| 2487 |
"mean": 0.23055555555555554,
|
| 2488 |
"n": 120,
|
| 2489 |
+
"num_samples": 1000
|
| 2490 |
},
|
| 2491 |
"mean": 0.23055555555555554,
|
| 2492 |
"per_seed": {
|
|
|
|
| 2619 |
"low": 0.07490909090909091,
|
| 2620 |
"mean": 0.18602020202020203,
|
| 2621 |
"n": 5,
|
| 2622 |
+
"num_samples": 1000
|
| 2623 |
},
|
| 2624 |
"mean": 0.18602020202020203,
|
| 2625 |
"per_task": {
|
|
|
|
| 2637 |
"low": 0.1527777777777778,
|
| 2638 |
"mean": 0.2222222222222222,
|
| 2639 |
"n": 144,
|
| 2640 |
+
"num_samples": 1000
|
| 2641 |
},
|
| 2642 |
"n": 144
|
| 2643 |
},
|
|
|
|
| 2645 |
"macro_by_seed": {
|
| 2646 |
"ci": {
|
| 2647 |
"confidence": 0.95,
|
| 2648 |
+
"high": 0.02430555555555555,
|
| 2649 |
+
"low": -0.14513888888888887,
|
| 2650 |
"mean": -0.061111111111111116,
|
| 2651 |
"n": 120,
|
| 2652 |
+
"num_samples": 1000
|
| 2653 |
},
|
| 2654 |
"mean": -0.061111111111111116,
|
| 2655 |
"per_seed": {
|
|
|
|
| 2778 |
"macro_by_task": {
|
| 2779 |
"ci": {
|
| 2780 |
"confidence": 0.95,
|
| 2781 |
+
"high": 0.06787878787878789,
|
| 2782 |
"low": -0.08412121212121212,
|
| 2783 |
"mean": -0.009454545454545455,
|
| 2784 |
"n": 5,
|
| 2785 |
+
"num_samples": 1000
|
| 2786 |
},
|
| 2787 |
"mean": -0.009454545454545455,
|
| 2788 |
"per_task": {
|
|
|
|
| 2797 |
"micro": {
|
| 2798 |
"confidence": 0.95,
|
| 2799 |
"high": 0.013888888888888888,
|
| 2800 |
+
"low": -0.14583333333333334,
|
| 2801 |
"mean": -0.0625,
|
| 2802 |
"n": 144,
|
| 2803 |
+
"num_samples": 1000
|
| 2804 |
},
|
| 2805 |
"n": 144
|
| 2806 |
},
|
|
|
|
| 2808 |
"macro_by_seed": {
|
| 2809 |
"ci": {
|
| 2810 |
"confidence": 0.95,
|
| 2811 |
+
"high": 0.7218117172195989,
|
| 2812 |
+
"low": 0.44103354020593943,
|
| 2813 |
"mean": 0.5765514451488142,
|
| 2814 |
"n": 120,
|
| 2815 |
+
"num_samples": 1000
|
| 2816 |
},
|
| 2817 |
"mean": 0.5765514451488142,
|
| 2818 |
"per_seed": {
|
|
|
|
| 2942 |
"ci": {
|
| 2943 |
"confidence": 0.95,
|
| 2944 |
"high": 0.7802921726552374,
|
| 2945 |
+
"low": 0.2568116229441431,
|
| 2946 |
"mean": 0.5089749424240053,
|
| 2947 |
"n": 5,
|
| 2948 |
+
"num_samples": 1000
|
| 2949 |
},
|
| 2950 |
"mean": 0.5089749424240053,
|
| 2951 |
"per_task": {
|
|
|
|
| 2959 |
"metric": "selected_utility_at_8",
|
| 2960 |
"micro": {
|
| 2961 |
"confidence": 0.95,
|
| 2962 |
+
"high": 0.6933383335223576,
|
| 2963 |
+
"low": 0.41932320297120085,
|
| 2964 |
"mean": 0.5485236065190596,
|
| 2965 |
"n": 144,
|
| 2966 |
+
"num_samples": 1000
|
| 2967 |
},
|
| 2968 |
"n": 144
|
| 2969 |
},
|
|
|
|
| 2971 |
"macro_by_seed": {
|
| 2972 |
"ci": {
|
| 2973 |
"confidence": 0.95,
|
| 2974 |
+
"high": -0.05383536980354822,
|
| 2975 |
+
"low": -0.3634254430313983,
|
| 2976 |
"mean": -0.20915491116207302,
|
| 2977 |
"n": 120,
|
| 2978 |
+
"num_samples": 1000
|
| 2979 |
},
|
| 2980 |
"mean": -0.20915491116207302,
|
| 2981 |
"per_seed": {
|
|
|
|
| 3104 |
"macro_by_task": {
|
| 3105 |
"ci": {
|
| 3106 |
"confidence": 0.95,
|
| 3107 |
+
"high": -0.03685359581280499,
|
| 3108 |
"low": -0.3532214832519548,
|
| 3109 |
"mean": -0.19503753953237987,
|
| 3110 |
"n": 5,
|
| 3111 |
+
"num_samples": 1000
|
| 3112 |
},
|
| 3113 |
"mean": -0.19503753953237987,
|
| 3114 |
"per_task": {
|
|
|
|
| 3122 |
"metric": "selected_utility_gain_over_base_at_8",
|
| 3123 |
"micro": {
|
| 3124 |
"confidence": 0.95,
|
| 3125 |
+
"high": -0.050540925846790846,
|
| 3126 |
"low": -0.34769073229431846,
|
| 3127 |
"mean": -0.1991940208495685,
|
| 3128 |
"n": 144,
|
| 3129 |
+
"num_samples": 1000
|
| 3130 |
},
|
| 3131 |
"n": 144
|
| 3132 |
},
|
|
|
|
| 3134 |
"macro_by_seed": {
|
| 3135 |
"ci": {
|
| 3136 |
"confidence": 0.95,
|
| 3137 |
+
"high": 0.7439352242830662,
|
| 3138 |
+
"low": 0.4503282577239184,
|
| 3139 |
"mean": 0.5889192757033419,
|
| 3140 |
"n": 120,
|
| 3141 |
+
"num_samples": 1000
|
| 3142 |
},
|
| 3143 |
"mean": 0.5889192757033419,
|
| 3144 |
"per_seed": {
|
|
|
|
| 3267 |
"macro_by_task": {
|
| 3268 |
"ci": {
|
| 3269 |
"confidence": 0.95,
|
| 3270 |
+
"high": 0.9107744794793879,
|
| 3271 |
+
"low": 0.30645303343608976,
|
| 3272 |
"mean": 0.5837108520707774,
|
| 3273 |
"n": 5,
|
| 3274 |
+
"num_samples": 1000
|
| 3275 |
},
|
| 3276 |
"mean": 0.5837108520707774,
|
| 3277 |
"per_task": {
|
|
|
|
| 3285 |
"metric": "selector_regret_at_8",
|
| 3286 |
"micro": {
|
| 3287 |
"confidence": 0.95,
|
| 3288 |
+
"high": 0.706521274279415,
|
| 3289 |
+
"low": 0.4344568868643061,
|
| 3290 |
"mean": 0.5604771331579994,
|
| 3291 |
"n": 144,
|
| 3292 |
+
"num_samples": 1000
|
| 3293 |
},
|
| 3294 |
"n": 144
|
| 3295 |
},
|
|
|
|
| 3297 |
"macro_by_seed": {
|
| 3298 |
"ci": {
|
| 3299 |
"confidence": 0.95,
|
| 3300 |
+
"high": 0.38333333333333336,
|
| 3301 |
+
"low": 0.2263888888888889,
|
| 3302 |
"mean": 0.3006944444444445,
|
| 3303 |
"n": 120,
|
| 3304 |
+
"num_samples": 1000
|
| 3305 |
},
|
| 3306 |
"mean": 0.3006944444444445,
|
| 3307 |
"per_seed": {
|
|
|
|
| 3430 |
"macro_by_task": {
|
| 3431 |
"ci": {
|
| 3432 |
"confidence": 0.95,
|
| 3433 |
+
"high": 0.4672727272727274,
|
| 3434 |
+
"low": 0.12000000000000002,
|
| 3435 |
"mean": 0.28949494949494947,
|
| 3436 |
"n": 5,
|
| 3437 |
+
"num_samples": 1000
|
| 3438 |
},
|
| 3439 |
"mean": 0.28949494949494947,
|
| 3440 |
"per_task": {
|
|
|
|
| 3449 |
"micro": {
|
| 3450 |
"confidence": 0.95,
|
| 3451 |
"high": 0.3611111111111111,
|
| 3452 |
+
"low": 0.2152777777777778,
|
| 3453 |
"mean": 0.2847222222222222,
|
| 3454 |
"n": 144,
|
| 3455 |
+
"num_samples": 1000
|
| 3456 |
},
|
| 3457 |
"n": 144
|
| 3458 |
},
|
|
|
|
| 3460 |
"macro_by_seed": {
|
| 3461 |
"ci": {
|
| 3462 |
"confidence": 0.95,
|
| 3463 |
+
"high": 0.37083333333333335,
|
| 3464 |
"low": 0.21111111111111114,
|
| 3465 |
"mean": 0.2916666666666667,
|
| 3466 |
"n": 120,
|
| 3467 |
+
"num_samples": 1000
|
| 3468 |
},
|
| 3469 |
"mean": 0.2916666666666667,
|
| 3470 |
"per_seed": {
|
|
|
|
| 3597 |
"low": 0.176,
|
| 3598 |
"mean": 0.43648484848484853,
|
| 3599 |
"n": 5,
|
| 3600 |
+
"num_samples": 1000
|
| 3601 |
},
|
| 3602 |
"mean": 0.43648484848484853,
|
| 3603 |
"per_task": {
|
|
|
|
| 3615 |
"low": 0.19444444444444445,
|
| 3616 |
"mean": 0.2638888888888889,
|
| 3617 |
"n": 144,
|
| 3618 |
+
"num_samples": 1000
|
| 3619 |
},
|
| 3620 |
"n": 144
|
| 3621 |
},
|
|
|
|
| 3623 |
"macro_by_seed": {
|
| 3624 |
"ci": {
|
| 3625 |
"confidence": 0.95,
|
| 3626 |
+
"high": 0.6333333333333333,
|
| 3627 |
+
"low": 0.47013888888888894,
|
| 3628 |
"mean": 0.554861111111111,
|
| 3629 |
"n": 120,
|
| 3630 |
+
"num_samples": 1000
|
| 3631 |
},
|
| 3632 |
"mean": 0.554861111111111,
|
| 3633 |
"per_seed": {
|
|
|
|
| 3760 |
"low": 0.457979797979798,
|
| 3761 |
"mean": 0.6846464646464646,
|
| 3762 |
"n": 5,
|
| 3763 |
+
"num_samples": 1000
|
| 3764 |
},
|
| 3765 |
"mean": 0.6846464646464646,
|
| 3766 |
"per_task": {
|
|
|
|
| 3778 |
"low": 0.4236111111111111,
|
| 3779 |
"mean": 0.5069444444444444,
|
| 3780 |
"n": 144,
|
| 3781 |
+
"num_samples": 1000
|
| 3782 |
},
|
| 3783 |
"n": 144
|
| 3784 |
},
|
|
|
|
| 3786 |
"macro_by_seed": {
|
| 3787 |
"ci": {
|
| 3788 |
"confidence": 0.95,
|
| 3789 |
+
"high": 0.6614446374736467,
|
| 3790 |
+
"low": 0.38915515649908533,
|
| 3791 |
"mean": 0.5202379634047651,
|
| 3792 |
"n": 120,
|
| 3793 |
+
"num_samples": 1000
|
| 3794 |
},
|
| 3795 |
"mean": 0.5202379634047651,
|
| 3796 |
"per_seed": {
|
|
|
|
| 3919 |
"macro_by_task": {
|
| 3920 |
"ci": {
|
| 3921 |
"confidence": 0.95,
|
| 3922 |
+
"high": 1.3049647501622788,
|
| 3923 |
"low": 0.32555825475768907,
|
| 3924 |
"mean": 0.7507728851216686,
|
| 3925 |
"n": 5,
|
| 3926 |
+
"num_samples": 1000
|
| 3927 |
},
|
| 3928 |
"mean": 0.7507728851216686,
|
| 3929 |
"per_task": {
|
|
|
|
| 3937 |
"metric": "support_gap_at_8",
|
| 3938 |
"micro": {
|
| 3939 |
"confidence": 0.95,
|
| 3940 |
+
"high": 0.6041254107549321,
|
| 3941 |
+
"low": 0.36498787499214,
|
| 3942 |
"mean": 0.48333957182411624,
|
| 3943 |
"n": 144,
|
| 3944 |
+
"num_samples": 1000
|
| 3945 |
},
|
| 3946 |
"n": 144
|
| 3947 |
},
|
|
|
|
| 3949 |
"macro_by_seed": {
|
| 3950 |
"ci": {
|
| 3951 |
"confidence": 0.95,
|
| 3952 |
+
"high": 1.1833753714913555,
|
| 3953 |
+
"low": 0.8840561589916532,
|
| 3954 |
"mean": 1.0425111003121452,
|
| 3955 |
"n": 120,
|
| 3956 |
+
"num_samples": 1000
|
| 3957 |
},
|
| 3958 |
"mean": 1.0425111003121452,
|
| 3959 |
"per_seed": {
|
|
|
|
| 4083 |
"ci": {
|
| 4084 |
"confidence": 0.95,
|
| 4085 |
"high": 1.6644046113617492,
|
| 4086 |
+
"low": 0.8802642338476849,
|
| 4087 |
"mean": 1.2606986467857102,
|
| 4088 |
"n": 5,
|
| 4089 |
+
"num_samples": 1000
|
| 4090 |
},
|
| 4091 |
"mean": 1.2606986467857102,
|
| 4092 |
"per_task": {
|
|
|
|
| 4101 |
"micro": {
|
| 4102 |
"confidence": 0.95,
|
| 4103 |
"high": 1.1140025374075373,
|
| 4104 |
+
"low": 0.8262412406863101,
|
| 4105 |
"mean": 0.9697925916213231,
|
| 4106 |
"n": 144,
|
| 4107 |
+
"num_samples": 1000
|
| 4108 |
},
|
| 4109 |
"n": 144
|
| 4110 |
}
|
workspace/runs/ctt_test_rollout_comparison/metrics_by_seed.json
CHANGED
|
The diff for this file is too large to render.
See raw diff
|
|
|
workspace/runs/ctt_test_rollout_comparison/metrics_by_task.json
CHANGED
|
@@ -1,16 +1,20 @@
|
|
| 1 |
{
|
| 2 |
"LiftPegUpright-v1": {
|
|
|
|
| 3 |
"base_success": 0.2,
|
| 4 |
"base_utility": 0.9498664379119873,
|
| 5 |
"branch_car_at_8": 0.5633442640304566,
|
|
|
|
| 6 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 7 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 8 |
"outcome_ptr_at_8": 0.7333333333333333,
|
| 9 |
"pairwise_causal_calibration_ece": 0.1621114782545992,
|
|
|
|
| 10 |
"proposal_oracle_success_at_8": 0.6666666666666666,
|
| 11 |
"proposal_oracle_success_gain_over_base_at_8": 0.4666666666666667,
|
| 12 |
"proposal_oracle_utility_at_8": 1.5397698879241943,
|
| 13 |
"proposal_oracle_utility_gain_over_base_at_8": 0.589903450012207,
|
|
|
|
| 14 |
"selected_success_at_8": 0.3333333333333333,
|
| 15 |
"selected_success_gain_over_base_at_8": 0.13333333333333333,
|
| 16 |
"selected_utility_at_8": 0.9764256238937378,
|
|
@@ -23,17 +27,21 @@
|
|
| 23 |
"total_car_to_hidden_at_8": 1.0235743761062621
|
| 24 |
},
|
| 25 |
"PickCube-v1": {
|
|
|
|
| 26 |
"base_success": 0.0,
|
| 27 |
"base_utility": 0.5839591026306152,
|
| 28 |
"branch_car_at_8": 0.15916230529546738,
|
|
|
|
| 29 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 30 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 31 |
"outcome_ptr_at_8": 0.0,
|
| 32 |
"pairwise_causal_calibration_ece": 0.2290622239534128,
|
|
|
|
| 33 |
"proposal_oracle_success_at_8": 0.0,
|
| 34 |
"proposal_oracle_success_gain_over_base_at_8": 0.0,
|
| 35 |
"proposal_oracle_utility_at_8": 0.25069022675355274,
|
| 36 |
"proposal_oracle_utility_gain_over_base_at_8": -0.3332688758770625,
|
|
|
|
| 37 |
"selected_success_at_8": 0.0,
|
| 38 |
"selected_success_gain_over_base_at_8": 0.0,
|
| 39 |
"selected_utility_at_8": 0.09152792145808537,
|
|
@@ -46,17 +54,21 @@
|
|
| 46 |
"total_car_to_hidden_at_8": 1.9084720785419147
|
| 47 |
},
|
| 48 |
"PullCube-v1": {
|
|
|
|
| 49 |
"base_success": 0.37333333333333335,
|
| 50 |
"base_utility": 0.7993177033324416,
|
| 51 |
"branch_car_at_8": 0.40464685219650465,
|
|
|
|
| 52 |
"hidden_chart_oracle_success_at_8": 0.52,
|
| 53 |
"hidden_chart_oracle_utility_at_8": 1.1467878860235214,
|
| 54 |
"outcome_ptr_at_8": 0.38666666666666666,
|
| 55 |
"pairwise_causal_calibration_ece": 0.17280987267234024,
|
|
|
|
| 56 |
"proposal_oracle_success_at_8": 0.4533333333333333,
|
| 57 |
"proposal_oracle_success_gain_over_base_at_8": 0.08,
|
| 58 |
"proposal_oracle_utility_at_8": 0.9498632888682187,
|
| 59 |
"proposal_oracle_utility_gain_over_base_at_8": 0.1505455855357771,
|
|
|
|
| 60 |
"selected_success_at_8": 0.25333333333333335,
|
| 61 |
"selected_success_gain_over_base_at_8": -0.12,
|
| 62 |
"selected_utility_at_8": 0.5452164366717139,
|
|
@@ -69,17 +81,21 @@
|
|
| 69 |
"total_car_to_hidden_at_8": 0.6015714493518074
|
| 70 |
},
|
| 71 |
"PushCube-v1": {
|
|
|
|
| 72 |
"base_success": 0.2222222222222222,
|
| 73 |
"base_utility": 0.49637971735662884,
|
| 74 |
"branch_car_at_8": 1.178912353184488,
|
|
|
|
| 75 |
"hidden_chart_oracle_success_at_8": 0.8333333333333334,
|
| 76 |
"hidden_chart_oracle_utility_at_8": 1.7015800600250561,
|
| 77 |
"outcome_ptr_at_8": 0.7777777777777778,
|
| 78 |
"pairwise_causal_calibration_ece": 0.1716092477675637,
|
|
|
|
| 79 |
"proposal_oracle_success_at_8": 0.8333333333333334,
|
| 80 |
"proposal_oracle_success_gain_over_base_at_8": 0.6111111111111112,
|
| 81 |
"proposal_oracle_utility_at_8": 1.6836495283577178,
|
| 82 |
"proposal_oracle_utility_gain_over_base_at_8": 1.1872698110010889,
|
|
|
|
| 83 |
"selected_success_at_8": 0.2222222222222222,
|
| 84 |
"selected_success_gain_over_base_at_8": 0.0,
|
| 85 |
"selected_utility_at_8": 0.5047371751732297,
|
|
@@ -92,17 +108,21 @@
|
|
| 92 |
"total_car_to_hidden_at_8": 1.1968428848518267
|
| 93 |
},
|
| 94 |
"StackCube-v1": {
|
|
|
|
| 95 |
"base_success": 0.18181818181818182,
|
| 96 |
"base_utility": 0.6905394485502532,
|
| 97 |
"branch_car_at_8": 0.6124884856469703,
|
|
|
|
| 98 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 99 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 100 |
"outcome_ptr_at_8": 0.6666666666666666,
|
| 101 |
"pairwise_causal_calibration_ece": 0.1741911827659335,
|
|
|
|
| 102 |
"proposal_oracle_success_at_8": 0.42424242424242425,
|
| 103 |
"proposal_oracle_success_gain_over_base_at_8": 0.24242424242424243,
|
| 104 |
"proposal_oracle_utility_at_8": 1.0394560405702302,
|
| 105 |
"proposal_oracle_utility_gain_over_base_at_8": 0.348916592019977,
|
|
|
|
| 106 |
"selected_success_at_8": 0.12121212121212122,
|
| 107 |
"selected_success_gain_over_base_at_8": -0.06060606060606061,
|
| 108 |
"selected_utility_at_8": 0.42696755492325983,
|
|
|
|
| 1 |
{
|
| 2 |
"LiftPegUpright-v1": {
|
| 3 |
+
"base_safety_label_known": 0.0,
|
| 4 |
"base_success": 0.2,
|
| 5 |
"base_utility": 0.9498664379119873,
|
| 6 |
"branch_car_at_8": 0.5633442640304566,
|
| 7 |
+
"generated_safety_label_coverage_at_8": 0.0,
|
| 8 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 9 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 10 |
"outcome_ptr_at_8": 0.7333333333333333,
|
| 11 |
"pairwise_causal_calibration_ece": 0.1621114782545992,
|
| 12 |
+
"proposal_oracle_safety_label_known_at_8": 0.0,
|
| 13 |
"proposal_oracle_success_at_8": 0.6666666666666666,
|
| 14 |
"proposal_oracle_success_gain_over_base_at_8": 0.4666666666666667,
|
| 15 |
"proposal_oracle_utility_at_8": 1.5397698879241943,
|
| 16 |
"proposal_oracle_utility_gain_over_base_at_8": 0.589903450012207,
|
| 17 |
+
"selected_safety_label_known_at_8": 0.0,
|
| 18 |
"selected_success_at_8": 0.3333333333333333,
|
| 19 |
"selected_success_gain_over_base_at_8": 0.13333333333333333,
|
| 20 |
"selected_utility_at_8": 0.9764256238937378,
|
|
|
|
| 27 |
"total_car_to_hidden_at_8": 1.0235743761062621
|
| 28 |
},
|
| 29 |
"PickCube-v1": {
|
| 30 |
+
"base_safety_label_known": 0.0,
|
| 31 |
"base_success": 0.0,
|
| 32 |
"base_utility": 0.5839591026306152,
|
| 33 |
"branch_car_at_8": 0.15916230529546738,
|
| 34 |
+
"generated_safety_label_coverage_at_8": 0.0,
|
| 35 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 36 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 37 |
"outcome_ptr_at_8": 0.0,
|
| 38 |
"pairwise_causal_calibration_ece": 0.2290622239534128,
|
| 39 |
+
"proposal_oracle_safety_label_known_at_8": 0.0,
|
| 40 |
"proposal_oracle_success_at_8": 0.0,
|
| 41 |
"proposal_oracle_success_gain_over_base_at_8": 0.0,
|
| 42 |
"proposal_oracle_utility_at_8": 0.25069022675355274,
|
| 43 |
"proposal_oracle_utility_gain_over_base_at_8": -0.3332688758770625,
|
| 44 |
+
"selected_safety_label_known_at_8": 0.0,
|
| 45 |
"selected_success_at_8": 0.0,
|
| 46 |
"selected_success_gain_over_base_at_8": 0.0,
|
| 47 |
"selected_utility_at_8": 0.09152792145808537,
|
|
|
|
| 54 |
"total_car_to_hidden_at_8": 1.9084720785419147
|
| 55 |
},
|
| 56 |
"PullCube-v1": {
|
| 57 |
+
"base_safety_label_known": 0.0,
|
| 58 |
"base_success": 0.37333333333333335,
|
| 59 |
"base_utility": 0.7993177033324416,
|
| 60 |
"branch_car_at_8": 0.40464685219650465,
|
| 61 |
+
"generated_safety_label_coverage_at_8": 0.0,
|
| 62 |
"hidden_chart_oracle_success_at_8": 0.52,
|
| 63 |
"hidden_chart_oracle_utility_at_8": 1.1467878860235214,
|
| 64 |
"outcome_ptr_at_8": 0.38666666666666666,
|
| 65 |
"pairwise_causal_calibration_ece": 0.17280987267234024,
|
| 66 |
+
"proposal_oracle_safety_label_known_at_8": 0.0,
|
| 67 |
"proposal_oracle_success_at_8": 0.4533333333333333,
|
| 68 |
"proposal_oracle_success_gain_over_base_at_8": 0.08,
|
| 69 |
"proposal_oracle_utility_at_8": 0.9498632888682187,
|
| 70 |
"proposal_oracle_utility_gain_over_base_at_8": 0.1505455855357771,
|
| 71 |
+
"selected_safety_label_known_at_8": 0.0,
|
| 72 |
"selected_success_at_8": 0.25333333333333335,
|
| 73 |
"selected_success_gain_over_base_at_8": -0.12,
|
| 74 |
"selected_utility_at_8": 0.5452164366717139,
|
|
|
|
| 81 |
"total_car_to_hidden_at_8": 0.6015714493518074
|
| 82 |
},
|
| 83 |
"PushCube-v1": {
|
| 84 |
+
"base_safety_label_known": 0.0,
|
| 85 |
"base_success": 0.2222222222222222,
|
| 86 |
"base_utility": 0.49637971735662884,
|
| 87 |
"branch_car_at_8": 1.178912353184488,
|
| 88 |
+
"generated_safety_label_coverage_at_8": 0.0,
|
| 89 |
"hidden_chart_oracle_success_at_8": 0.8333333333333334,
|
| 90 |
"hidden_chart_oracle_utility_at_8": 1.7015800600250561,
|
| 91 |
"outcome_ptr_at_8": 0.7777777777777778,
|
| 92 |
"pairwise_causal_calibration_ece": 0.1716092477675637,
|
| 93 |
+
"proposal_oracle_safety_label_known_at_8": 0.0,
|
| 94 |
"proposal_oracle_success_at_8": 0.8333333333333334,
|
| 95 |
"proposal_oracle_success_gain_over_base_at_8": 0.6111111111111112,
|
| 96 |
"proposal_oracle_utility_at_8": 1.6836495283577178,
|
| 97 |
"proposal_oracle_utility_gain_over_base_at_8": 1.1872698110010889,
|
| 98 |
+
"selected_safety_label_known_at_8": 0.0,
|
| 99 |
"selected_success_at_8": 0.2222222222222222,
|
| 100 |
"selected_success_gain_over_base_at_8": 0.0,
|
| 101 |
"selected_utility_at_8": 0.5047371751732297,
|
|
|
|
| 108 |
"total_car_to_hidden_at_8": 1.1968428848518267
|
| 109 |
},
|
| 110 |
"StackCube-v1": {
|
| 111 |
+
"base_safety_label_known": 0.0,
|
| 112 |
"base_success": 0.18181818181818182,
|
| 113 |
"base_utility": 0.6905394485502532,
|
| 114 |
"branch_car_at_8": 0.6124884856469703,
|
| 115 |
+
"generated_safety_label_coverage_at_8": 0.0,
|
| 116 |
"hidden_chart_oracle_success_at_8": 1.0,
|
| 117 |
"hidden_chart_oracle_utility_at_8": 2.0,
|
| 118 |
"outcome_ptr_at_8": 0.6666666666666666,
|
| 119 |
"pairwise_causal_calibration_ece": 0.1741911827659335,
|
| 120 |
+
"proposal_oracle_safety_label_known_at_8": 0.0,
|
| 121 |
"proposal_oracle_success_at_8": 0.42424242424242425,
|
| 122 |
"proposal_oracle_success_gain_over_base_at_8": 0.24242424242424243,
|
| 123 |
"proposal_oracle_utility_at_8": 1.0394560405702302,
|
| 124 |
"proposal_oracle_utility_gain_over_base_at_8": 0.348916592019977,
|
| 125 |
+
"selected_safety_label_known_at_8": 0.0,
|
| 126 |
"selected_success_at_8": 0.12121212121212122,
|
| 127 |
"selected_success_gain_over_base_at_8": -0.06060606060606061,
|
| 128 |
"selected_utility_at_8": 0.42696755492325983,
|
workspace/runs/ctt_test_rollout_comparison/report.md
CHANGED
|
@@ -6,27 +6,31 @@ K: `8`
|
|
| 6 |
|
| 7 |
| Metric | N | Micro mean | 95% CI |
|
| 8 |
| --- | ---: | ---: | ---: |
|
|
|
|
| 9 |
| base_success | 144 | 0.2847 | [0.2153, 0.3542] |
|
| 10 |
-
| base_utility | 144 | 0.7477 | [0.
|
| 11 |
-
| branch_car_at_8 | 144 | 0.5605 | [0.
|
| 12 |
-
|
|
| 13 |
-
|
|
| 14 |
-
|
|
| 15 |
-
|
|
| 16 |
-
|
|
| 17 |
-
|
|
| 18 |
-
|
|
| 19 |
-
|
|
|
|
|
|
|
|
|
|
|
| 20 |
| selected_success_at_8 | 144 | 0.2222 | [0.1528, 0.2917] |
|
| 21 |
-
| selected_success_gain_over_base_at_8 | 144 | -0.0625 | [-0.
|
| 22 |
-
| selected_utility_at_8 | 144 | 0.5485 | [0.
|
| 23 |
-
| selected_utility_gain_over_base_at_8 | 144 | -0.1992 | [-0.3477, -0.
|
| 24 |
-
| selector_regret_at_8 | 144 | 0.5605 | [0.
|
| 25 |
-
| success_selector_gap_at_8 | 144 | 0.2847 | [0.
|
| 26 |
| success_support_gap_at_8 | 144 | 0.2639 | [0.1944, 0.3333] |
|
| 27 |
| success_total_car_to_hidden_at_8 | 144 | 0.5069 | [0.4236, 0.5903] |
|
| 28 |
-
| support_gap_at_8 | 144 | 0.4833 | [0.
|
| 29 |
-
| total_car_to_hidden_at_8 | 144 | 0.9698 | [0.
|
| 30 |
|
| 31 |
| Success/Utility | Mean |
|
| 32 |
| --- | ---: |
|
|
@@ -42,6 +46,15 @@ K: `8`
|
|
| 42 |
| selected_utility_mean | 0.5485 |
|
| 43 |
| proposal_oracle_utility_mean | 1.1090 |
|
| 44 |
| hidden_chart_oracle_utility_mean | 1.5183 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
|
| 46 |
These are measured generated-candidate rollouts, not PPTC proxies.
|
| 47 |
|
|
|
|
| 6 |
|
| 7 |
| Metric | N | Micro mean | 95% CI |
|
| 8 |
| --- | ---: | ---: | ---: |
|
| 9 |
+
| base_safety_label_known | 144 | 0.0000 | [0.0000, 0.0000] |
|
| 10 |
| base_success | 144 | 0.2847 | [0.2153, 0.3542] |
|
| 11 |
+
| base_utility | 144 | 0.7477 | [0.6173, 0.8788] |
|
| 12 |
+
| branch_car_at_8 | 144 | 0.5605 | [0.4345, 0.7065] |
|
| 13 |
+
| generated_safety_label_coverage_at_8 | 144 | 0.0000 | [0.0000, 0.0000] |
|
| 14 |
+
| hidden_chart_oracle_success_at_8 | 144 | 0.7292 | [0.6597, 0.7986] |
|
| 15 |
+
| hidden_chart_oracle_utility_at_8 | 144 | 1.5183 | [1.3941, 1.6438] |
|
| 16 |
+
| outcome_ptr_at_8 | 144 | 0.5278 | [0.4514, 0.6042] |
|
| 17 |
+
| pairwise_causal_calibration_ece | 144 | 0.1730 | [0.1547, 0.1911] |
|
| 18 |
+
| proposal_oracle_safety_label_known_at_8 | 144 | 0.0000 | [0.0000, 0.0000] |
|
| 19 |
+
| proposal_oracle_success_at_8 | 144 | 0.5069 | [0.4306, 0.5903] |
|
| 20 |
+
| proposal_oracle_success_gain_over_base_at_8 | 144 | 0.2222 | [0.1319, 0.3056] |
|
| 21 |
+
| proposal_oracle_utility_at_8 | 144 | 1.1090 | [0.9685, 1.2662] |
|
| 22 |
+
| proposal_oracle_utility_gain_over_base_at_8 | 144 | 0.3613 | [0.2109, 0.5091] |
|
| 23 |
+
| selected_safety_label_known_at_8 | 144 | 0.0000 | [0.0000, 0.0000] |
|
| 24 |
| selected_success_at_8 | 144 | 0.2222 | [0.1528, 0.2917] |
|
| 25 |
+
| selected_success_gain_over_base_at_8 | 144 | -0.0625 | [-0.1458, 0.0139] |
|
| 26 |
+
| selected_utility_at_8 | 144 | 0.5485 | [0.4193, 0.6933] |
|
| 27 |
+
| selected_utility_gain_over_base_at_8 | 144 | -0.1992 | [-0.3477, -0.0505] |
|
| 28 |
+
| selector_regret_at_8 | 144 | 0.5605 | [0.4345, 0.7065] |
|
| 29 |
+
| success_selector_gap_at_8 | 144 | 0.2847 | [0.2153, 0.3611] |
|
| 30 |
| success_support_gap_at_8 | 144 | 0.2639 | [0.1944, 0.3333] |
|
| 31 |
| success_total_car_to_hidden_at_8 | 144 | 0.5069 | [0.4236, 0.5903] |
|
| 32 |
+
| support_gap_at_8 | 144 | 0.4833 | [0.3650, 0.6041] |
|
| 33 |
+
| total_car_to_hidden_at_8 | 144 | 0.9698 | [0.8262, 1.1140] |
|
| 34 |
|
| 35 |
| Success/Utility | Mean |
|
| 36 |
| --- | ---: |
|
|
|
|
| 46 |
| selected_utility_mean | 0.5485 |
|
| 47 |
| proposal_oracle_utility_mean | 1.1090 |
|
| 48 |
| hidden_chart_oracle_utility_mean | 1.5183 |
|
| 49 |
+
| generated_safety_label_coverage | 0.0000 |
|
| 50 |
+
| generated_unsafe_rate_known | n/a |
|
| 51 |
+
| any_generated_unsafe_known | n/a |
|
| 52 |
+
| selected_safety_label_known_rate | 0.0000 |
|
| 53 |
+
| selected_unsafe_rate_known | n/a |
|
| 54 |
+
| proposal_oracle_safety_label_known_rate | 0.0000 |
|
| 55 |
+
| proposal_oracle_unsafe_rate_known | n/a |
|
| 56 |
+
| base_safety_label_known_rate | 0.0000 |
|
| 57 |
+
| base_unsafe_rate_known | n/a |
|
| 58 |
|
| 59 |
These are measured generated-candidate rollouts, not PPTC proxies.
|
| 60 |
|
workspace/runs/ctt_test_rollout_comparison/table.tex
CHANGED
|
@@ -3,26 +3,30 @@
|
|
| 3 |
\toprule
|
| 4 |
Metric & N & Micro mean & CI low & CI high \\
|
| 5 |
\midrule
|
|
|
|
| 6 |
base\_success & 144 & 0.2847 & 0.2153 & 0.3542 \\
|
| 7 |
-
base\_utility & 144 & 0.7477 & 0.
|
| 8 |
-
branch\_car\_at\_8 & 144 & 0.5605 & 0.
|
| 9 |
-
|
| 10 |
-
hidden\_chart\_oracle\
|
| 11 |
-
|
| 12 |
-
|
| 13 |
-
|
| 14 |
-
proposal\_oracle\
|
| 15 |
-
proposal\_oracle\
|
| 16 |
-
proposal\_oracle\
|
|
|
|
|
|
|
|
|
|
| 17 |
selected\_success\_at\_8 & 144 & 0.2222 & 0.1528 & 0.2917 \\
|
| 18 |
-
selected\_success\_gain\_over\_base\_at\_8 & 144 & -0.0625 & -0.
|
| 19 |
-
selected\_utility\_at\_8 & 144 & 0.5485 & 0.
|
| 20 |
-
selected\_utility\_gain\_over\_base\_at\_8 & 144 & -0.1992 & -0.3477 & -0.
|
| 21 |
-
selector\_regret\_at\_8 & 144 & 0.5605 & 0.
|
| 22 |
-
success\_selector\_gap\_at\_8 & 144 & 0.2847 & 0.
|
| 23 |
success\_support\_gap\_at\_8 & 144 & 0.2639 & 0.1944 & 0.3333 \\
|
| 24 |
success\_total\_car\_to\_hidden\_at\_8 & 144 & 0.5069 & 0.4236 & 0.5903 \\
|
| 25 |
-
support\_gap\_at\_8 & 144 & 0.4833 & 0.
|
| 26 |
-
total\_car\_to\_hidden\_at\_8 & 144 & 0.9698 & 0.
|
| 27 |
\bottomrule
|
| 28 |
\end{tabular}
|
|
|
|
| 3 |
\toprule
|
| 4 |
Metric & N & Micro mean & CI low & CI high \\
|
| 5 |
\midrule
|
| 6 |
+
base\_safety\_label\_known & 144 & 0.0000 & 0.0000 & 0.0000 \\
|
| 7 |
base\_success & 144 & 0.2847 & 0.2153 & 0.3542 \\
|
| 8 |
+
base\_utility & 144 & 0.7477 & 0.6173 & 0.8788 \\
|
| 9 |
+
branch\_car\_at\_8 & 144 & 0.5605 & 0.4345 & 0.7065 \\
|
| 10 |
+
generated\_safety\_label\_coverage\_at\_8 & 144 & 0.0000 & 0.0000 & 0.0000 \\
|
| 11 |
+
hidden\_chart\_oracle\_success\_at\_8 & 144 & 0.7292 & 0.6597 & 0.7986 \\
|
| 12 |
+
hidden\_chart\_oracle\_utility\_at\_8 & 144 & 1.5183 & 1.3941 & 1.6438 \\
|
| 13 |
+
outcome\_ptr\_at\_8 & 144 & 0.5278 & 0.4514 & 0.6042 \\
|
| 14 |
+
pairwise\_causal\_calibration\_ece & 144 & 0.1730 & 0.1547 & 0.1911 \\
|
| 15 |
+
proposal\_oracle\_safety\_label\_known\_at\_8 & 144 & 0.0000 & 0.0000 & 0.0000 \\
|
| 16 |
+
proposal\_oracle\_success\_at\_8 & 144 & 0.5069 & 0.4306 & 0.5903 \\
|
| 17 |
+
proposal\_oracle\_success\_gain\_over\_base\_at\_8 & 144 & 0.2222 & 0.1319 & 0.3056 \\
|
| 18 |
+
proposal\_oracle\_utility\_at\_8 & 144 & 1.1090 & 0.9685 & 1.2662 \\
|
| 19 |
+
proposal\_oracle\_utility\_gain\_over\_base\_at\_8 & 144 & 0.3613 & 0.2109 & 0.5091 \\
|
| 20 |
+
selected\_safety\_label\_known\_at\_8 & 144 & 0.0000 & 0.0000 & 0.0000 \\
|
| 21 |
selected\_success\_at\_8 & 144 & 0.2222 & 0.1528 & 0.2917 \\
|
| 22 |
+
selected\_success\_gain\_over\_base\_at\_8 & 144 & -0.0625 & -0.1458 & 0.0139 \\
|
| 23 |
+
selected\_utility\_at\_8 & 144 & 0.5485 & 0.4193 & 0.6933 \\
|
| 24 |
+
selected\_utility\_gain\_over\_base\_at\_8 & 144 & -0.1992 & -0.3477 & -0.0505 \\
|
| 25 |
+
selector\_regret\_at\_8 & 144 & 0.5605 & 0.4345 & 0.7065 \\
|
| 26 |
+
success\_selector\_gap\_at\_8 & 144 & 0.2847 & 0.2153 & 0.3611 \\
|
| 27 |
success\_support\_gap\_at\_8 & 144 & 0.2639 & 0.1944 & 0.3333 \\
|
| 28 |
success\_total\_car\_to\_hidden\_at\_8 & 144 & 0.5069 & 0.4236 & 0.5903 \\
|
| 29 |
+
support\_gap\_at\_8 & 144 & 0.4833 & 0.3650 & 0.6041 \\
|
| 30 |
+
total\_car\_to\_hidden\_at\_8 & 144 & 0.9698 & 0.8262 & 1.1140 \\
|
| 31 |
\bottomrule
|
| 32 |
\end{tabular}
|
workspace/runs/ctt_val_proxy_comparison/eval.log
CHANGED
|
@@ -3,4 +3,6 @@ task_memory: runs/task_memory_val_proxy
|
|
| 3 |
ctt_residual_full: runs/ctt_residual_full_seed0_val_proxy,runs/ctt_residual_full_seed1_val_proxy,runs/ctt_residual_full_seed2_val_proxy
|
| 4 |
ctt_residual_base_context: runs/ctt_residual_base_context_seed0_val_proxy
|
| 5 |
ctt_residual_base_context_obs: runs/ctt_residual_base_context_obs_seed0_val_proxy,runs/ctt_residual_base_context_obs_seed1_val_proxy,runs/ctt_residual_base_context_obs_seed2_val_proxy
|
|
|
|
|
|
|
| 6 |
ctt_gated_residual_full: runs/ctt_gated_residual_full_seed0_val_proxy,runs/ctt_gated_residual_full_seed1_val_proxy,runs/ctt_gated_residual_full_seed2_val_proxy
|
|
|
|
| 3 |
ctt_residual_full: runs/ctt_residual_full_seed0_val_proxy,runs/ctt_residual_full_seed1_val_proxy,runs/ctt_residual_full_seed2_val_proxy
|
| 4 |
ctt_residual_base_context: runs/ctt_residual_base_context_seed0_val_proxy
|
| 5 |
ctt_residual_base_context_obs: runs/ctt_residual_base_context_obs_seed0_val_proxy,runs/ctt_residual_base_context_obs_seed1_val_proxy,runs/ctt_residual_base_context_obs_seed2_val_proxy
|
| 6 |
+
ctt_residual_base_context_obj: runs/ctt_residual_base_context_obj_seed0_val_proxy,runs/ctt_residual_base_context_obj_seed1_val_proxy,runs/ctt_residual_base_context_obj_seed2_val_proxy
|
| 7 |
+
ctt_residual_base_context_obs_obj: runs/ctt_residual_base_context_obs_obj_seed0_val_proxy,runs/ctt_residual_base_context_obs_obj_seed1_val_proxy,runs/ctt_residual_base_context_obs_obj_seed2_val_proxy
|
| 8 |
ctt_gated_residual_full: runs/ctt_gated_residual_full_seed0_val_proxy,runs/ctt_gated_residual_full_seed1_val_proxy,runs/ctt_gated_residual_full_seed2_val_proxy
|
workspace/runs/ctt_val_proxy_comparison/metrics.json
CHANGED
|
@@ -93,6 +93,42 @@
|
|
| 93 |
"split_hash": "9d1dc1c10868f76fa080895df0f2730c173f12b6db14932599d55c2b60859381",
|
| 94 |
"train_seeds": 3
|
| 95 |
},
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 96 |
{
|
| 97 |
"beats_local_atlas_support": true,
|
| 98 |
"candidate_diversity": 0.1163743674776266,
|
|
|
|
| 93 |
"split_hash": "9d1dc1c10868f76fa080895df0f2730c173f12b6db14932599d55c2b60859381",
|
| 94 |
"train_seeds": 3
|
| 95 |
},
|
| 96 |
+
{
|
| 97 |
+
"beats_local_atlas_support": true,
|
| 98 |
+
"candidate_diversity": 0.24174490402345686,
|
| 99 |
+
"data_hash": "67f7b4a692f7c7e71da24378dc71e3399c5d35afc14c494869eba6087b765c42",
|
| 100 |
+
"mean_positive_distance": 0.43402743826233176,
|
| 101 |
+
"method": "ctt_residual_base_context_obj",
|
| 102 |
+
"negative_near_0p20": 0.038024455415759766,
|
| 103 |
+
"negative_near_0p40": 0.30944174585478934,
|
| 104 |
+
"num_rows": 207,
|
| 105 |
+
"pptc_0p20": 0.22705314009661837,
|
| 106 |
+
"pptc_0p40": 0.6425120772946861,
|
| 107 |
+
"proxy_gate_pass": true,
|
| 108 |
+
"proxy_support_distance": 0.3603756988567814,
|
| 109 |
+
"run_path": "runs/ctt_residual_base_context_obj_seed0_val_proxy,runs/ctt_residual_base_context_obj_seed1_val_proxy,runs/ctt_residual_base_context_obj_seed2_val_proxy",
|
| 110 |
+
"safety_ok_vs_local_atlas": true,
|
| 111 |
+
"split_hash": "9d1dc1c10868f76fa080895df0f2730c173f12b6db14932599d55c2b60859381",
|
| 112 |
+
"train_seeds": 3
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"beats_local_atlas_support": true,
|
| 116 |
+
"candidate_diversity": 0.2447676812446944,
|
| 117 |
+
"data_hash": "67f7b4a692f7c7e71da24378dc71e3399c5d35afc14c494869eba6087b765c42",
|
| 118 |
+
"mean_positive_distance": 0.4429413077106721,
|
| 119 |
+
"method": "ctt_residual_base_context_obs_obj",
|
| 120 |
+
"negative_near_0p20": 0.02853647962343615,
|
| 121 |
+
"negative_near_0p40": 0.2934249526640831,
|
| 122 |
+
"num_rows": 207,
|
| 123 |
+
"pptc_0p20": 0.20772946859903382,
|
| 124 |
+
"pptc_0p40": 0.642512077294686,
|
| 125 |
+
"proxy_gate_pass": true,
|
| 126 |
+
"proxy_support_distance": 0.3673834324291745,
|
| 127 |
+
"run_path": "runs/ctt_residual_base_context_obs_obj_seed0_val_proxy,runs/ctt_residual_base_context_obs_obj_seed1_val_proxy,runs/ctt_residual_base_context_obs_obj_seed2_val_proxy",
|
| 128 |
+
"safety_ok_vs_local_atlas": true,
|
| 129 |
+
"split_hash": "9d1dc1c10868f76fa080895df0f2730c173f12b6db14932599d55c2b60859381",
|
| 130 |
+
"train_seeds": 3
|
| 131 |
+
},
|
| 132 |
{
|
| 133 |
"beats_local_atlas_support": true,
|
| 134 |
"candidate_diversity": 0.1163743674776266,
|
workspace/runs/ctt_val_proxy_comparison/report.md
CHANGED
|
@@ -9,6 +9,8 @@ Safety slack over local-atlas NegativeNear@0.20: `0.0100`
|
|
| 9 |
| ctt_residual_full | 3 | 207 | 0.1981 | 0.6087 | 0.0296 | 0.4509 | 0.2703 | pass | `runs/ctt_residual_full_seed0_val_proxy,runs/ctt_residual_full_seed1_val_proxy,runs/ctt_residual_full_seed2_val_proxy` |
|
| 10 |
| ctt_residual_base_context | 1 | 69 | 0.1739 | 0.6232 | 0.0182 | 0.4429 | 0.2471 | pass | `runs/ctt_residual_base_context_seed0_val_proxy` |
|
| 11 |
| ctt_residual_base_context_obs | 3 | 207 | 0.2464 | 0.6425 | 0.0343 | 0.4347 | 0.2397 | pass | `runs/ctt_residual_base_context_obs_seed0_val_proxy,runs/ctt_residual_base_context_obs_seed1_val_proxy,runs/ctt_residual_base_context_obs_seed2_val_proxy` |
|
|
|
|
|
|
|
| 12 |
| ctt_gated_residual_full | 3 | 207 | 0.2319 | 0.6135 | 0.0527 | 0.4337 | 0.1164 | fail | `runs/ctt_gated_residual_full_seed0_val_proxy,runs/ctt_gated_residual_full_seed1_val_proxy,runs/ctt_gated_residual_full_seed2_val_proxy` |
|
| 13 |
|
| 14 |
Gate is proxy-only. It is not OutcomePTR or measured rollout success.
|
|
|
|
| 9 |
| ctt_residual_full | 3 | 207 | 0.1981 | 0.6087 | 0.0296 | 0.4509 | 0.2703 | pass | `runs/ctt_residual_full_seed0_val_proxy,runs/ctt_residual_full_seed1_val_proxy,runs/ctt_residual_full_seed2_val_proxy` |
|
| 10 |
| ctt_residual_base_context | 1 | 69 | 0.1739 | 0.6232 | 0.0182 | 0.4429 | 0.2471 | pass | `runs/ctt_residual_base_context_seed0_val_proxy` |
|
| 11 |
| ctt_residual_base_context_obs | 3 | 207 | 0.2464 | 0.6425 | 0.0343 | 0.4347 | 0.2397 | pass | `runs/ctt_residual_base_context_obs_seed0_val_proxy,runs/ctt_residual_base_context_obs_seed1_val_proxy,runs/ctt_residual_base_context_obs_seed2_val_proxy` |
|
| 12 |
+
| ctt_residual_base_context_obj | 3 | 207 | 0.2271 | 0.6425 | 0.0380 | 0.4340 | 0.2417 | pass | `runs/ctt_residual_base_context_obj_seed0_val_proxy,runs/ctt_residual_base_context_obj_seed1_val_proxy,runs/ctt_residual_base_context_obj_seed2_val_proxy` |
|
| 13 |
+
| ctt_residual_base_context_obs_obj | 3 | 207 | 0.2077 | 0.6425 | 0.0285 | 0.4429 | 0.2448 | pass | `runs/ctt_residual_base_context_obs_obj_seed0_val_proxy,runs/ctt_residual_base_context_obs_obj_seed1_val_proxy,runs/ctt_residual_base_context_obs_obj_seed2_val_proxy` |
|
| 14 |
| ctt_gated_residual_full | 3 | 207 | 0.2319 | 0.6135 | 0.0527 | 0.4337 | 0.1164 | fail | `runs/ctt_gated_residual_full_seed0_val_proxy,runs/ctt_gated_residual_full_seed1_val_proxy,runs/ctt_gated_residual_full_seed2_val_proxy` |
|
| 15 |
|
| 16 |
Gate is proxy-only. It is not OutcomePTR or measured rollout success.
|
workspace/runs/ctt_val_proxy_comparison/table.tex
CHANGED
|
@@ -8,6 +8,8 @@ task\_memory & 1 & 69 & 0.3188 & 0.5362 & 0.0175 & 0.9616 & 0.9751 & baseline \\
|
|
| 8 |
ctt\_residual\_full & 3 & 207 & 0.1981 & 0.6087 & 0.0296 & 0.4509 & 0.2703 & pass \\
|
| 9 |
ctt\_residual\_base\_context & 1 & 69 & 0.1739 & 0.6232 & 0.0182 & 0.4429 & 0.2471 & pass \\
|
| 10 |
ctt\_residual\_base\_context\_obs & 3 & 207 & 0.2464 & 0.6425 & 0.0343 & 0.4347 & 0.2397 & pass \\
|
|
|
|
|
|
|
| 11 |
ctt\_gated\_residual\_full & 3 & 207 & 0.2319 & 0.6135 & 0.0527 & 0.4337 & 0.1164 & fail \\
|
| 12 |
\bottomrule
|
| 13 |
\end{tabular}
|
|
|
|
| 8 |
ctt\_residual\_full & 3 & 207 & 0.1981 & 0.6087 & 0.0296 & 0.4509 & 0.2703 & pass \\
|
| 9 |
ctt\_residual\_base\_context & 1 & 69 & 0.1739 & 0.6232 & 0.0182 & 0.4429 & 0.2471 & pass \\
|
| 10 |
ctt\_residual\_base\_context\_obs & 3 & 207 & 0.2464 & 0.6425 & 0.0343 & 0.4347 & 0.2397 & pass \\
|
| 11 |
+
ctt\_residual\_base\_context\_obj & 3 & 207 & 0.2271 & 0.6425 & 0.0380 & 0.4340 & 0.2417 & pass \\
|
| 12 |
+
ctt\_residual\_base\_context\_obs\_obj & 3 & 207 & 0.2077 & 0.6425 & 0.0285 & 0.4429 & 0.2448 & pass \\
|
| 13 |
ctt\_gated\_residual\_full & 3 & 207 & 0.2319 & 0.6135 & 0.0527 & 0.4337 & 0.1164 & fail \\
|
| 14 |
\bottomrule
|
| 15 |
\end{tabular}
|
workspace/runs/ctt_val_rollout_comparison/combined_measured_candidates.json
CHANGED
|
@@ -1,4 +1,5 @@
|
|
| 1 |
{
|
|
|
|
| 2 |
"k": 8,
|
| 3 |
"num_rows": 207,
|
| 4 |
"report_type": "ctt_measured_rollout_comparison",
|
|
@@ -112358,6 +112359,7 @@
|
|
| 112358 |
],
|
| 112359 |
"schema_version": 1,
|
| 112360 |
"source_content_hash": "90b431047a1f35cec095cbafa7de5bc0126cb0d0f4db3ff31843dca369977957",
|
|
|
|
| 112361 |
"target_content_hash": "fa0fe0e3881ee70ce91492445871614d01b3a47fc7c576659c53ff17ddee4e42",
|
| 112362 |
"target_split_hash": "9d1dc1c10868f76fa080895df0f2730c173f12b6db14932599d55c2b60859381",
|
| 112363 |
"train_seeds": [
|
|
|
|
| 1 |
{
|
| 2 |
+
"data_hash": "fa0fe0e3881ee70ce91492445871614d01b3a47fc7c576659c53ff17ddee4e42",
|
| 3 |
"k": 8,
|
| 4 |
"num_rows": 207,
|
| 5 |
"report_type": "ctt_measured_rollout_comparison",
|
|
|
|
| 112359 |
],
|
| 112360 |
"schema_version": 1,
|
| 112361 |
"source_content_hash": "90b431047a1f35cec095cbafa7de5bc0126cb0d0f4db3ff31843dca369977957",
|
| 112362 |
+
"split_hash": "9d1dc1c10868f76fa080895df0f2730c173f12b6db14932599d55c2b60859381",
|
| 112363 |
"target_content_hash": "fa0fe0e3881ee70ce91492445871614d01b3a47fc7c576659c53ff17ddee4e42",
|
| 112364 |
"target_split_hash": "9d1dc1c10868f76fa080895df0f2730c173f12b6db14932599d55c2b60859381",
|
| 112365 |
"train_seeds": [
|
workspace/runs/ctt_val_rollout_comparison/command.txt
CHANGED
|
@@ -1 +1 @@
|
|
| 1 |
-
python scripts/build_ctt_rollout_comparison.py --run-glob runs/ctt_residual_rollout_val69_seed* --out-dir runs/ctt_val_rollout_comparison --k 8
|
|
|
|
| 1 |
+
python scripts/build_ctt_rollout_comparison.py --run-glob runs/ctt_residual_rollout_val69_seed* --out-dir runs/ctt_val_rollout_comparison --k 8
|
workspace/runs/ctt_val_rollout_comparison/measured_metrics/metrics.json
CHANGED
|
@@ -5599,11 +5599,11 @@
|
|
| 5599 |
"macro_by_seed": {
|
| 5600 |
"ci": {
|
| 5601 |
"confidence": 0.95,
|
| 5602 |
-
"high": 0.
|
| 5603 |
-
"low": 0.
|
| 5604 |
"mean": 0.322695035460993,
|
| 5605 |
"n": 141,
|
| 5606 |
-
"num_samples":
|
| 5607 |
},
|
| 5608 |
"mean": 0.322695035460993,
|
| 5609 |
"per_seed": {
|
|
@@ -5757,7 +5757,7 @@
|
|
| 5757 |
"low": 0.10666666666666666,
|
| 5758 |
"mean": 0.23728506787330317,
|
| 5759 |
"n": 5,
|
| 5760 |
-
"num_samples":
|
| 5761 |
},
|
| 5762 |
"mean": 0.23728506787330317,
|
| 5763 |
"per_task": {
|
|
@@ -5775,7 +5775,7 @@
|
|
| 5775 |
"low": 0.22705314009661837,
|
| 5776 |
"mean": 0.28502415458937197,
|
| 5777 |
"n": 207,
|
| 5778 |
-
"num_samples":
|
| 5779 |
},
|
| 5780 |
"n": 207
|
| 5781 |
},
|
|
@@ -5783,11 +5783,11 @@
|
|
| 5783 |
"macro_by_seed": {
|
| 5784 |
"ci": {
|
| 5785 |
"confidence": 0.95,
|
| 5786 |
-
"high": 0.
|
| 5787 |
-
"low": 0.
|
| 5788 |
"mean": 0.84238802050422,
|
| 5789 |
"n": 141,
|
| 5790 |
-
"num_samples":
|
| 5791 |
},
|
| 5792 |
"mean": 0.84238802050422,
|
| 5793 |
"per_seed": {
|
|
@@ -5937,11 +5937,11 @@
|
|
| 5937 |
"macro_by_task": {
|
| 5938 |
"ci": {
|
| 5939 |
"confidence": 0.95,
|
| 5940 |
-
"high": 0.
|
| 5941 |
-
"low": 0.
|
| 5942 |
"mean": 0.6891968629765195,
|
| 5943 |
"n": 5,
|
| 5944 |
-
"num_samples":
|
| 5945 |
},
|
| 5946 |
"mean": 0.6891968629765195,
|
| 5947 |
"per_task": {
|
|
@@ -5956,10 +5956,10 @@
|
|
| 5956 |
"micro": {
|
| 5957 |
"confidence": 0.95,
|
| 5958 |
"high": 0.851249745030154,
|
| 5959 |
-
"low": 0.
|
| 5960 |
"mean": 0.7358034186915535,
|
| 5961 |
"n": 207,
|
| 5962 |
-
"num_samples":
|
| 5963 |
},
|
| 5964 |
"n": 207
|
| 5965 |
},
|
|
@@ -5967,11 +5967,11 @@
|
|
| 5967 |
"macro_by_seed": {
|
| 5968 |
"ci": {
|
| 5969 |
"confidence": 0.95,
|
| 5970 |
-
"high": 0.
|
| 5971 |
-
"low": 0.
|
| 5972 |
"mean": 0.40561739670197494,
|
| 5973 |
"n": 141,
|
| 5974 |
-
"num_samples":
|
| 5975 |
},
|
| 5976 |
"mean": 0.40561739670197494,
|
| 5977 |
"per_seed": {
|
|
@@ -6121,11 +6121,11 @@
|
|
| 6121 |
"macro_by_task": {
|
| 6122 |
"ci": {
|
| 6123 |
"confidence": 0.95,
|
| 6124 |
-
"high": 0.
|
| 6125 |
-
"low": 0.
|
| 6126 |
"mean": 0.3798908010369334,
|
| 6127 |
"n": 5,
|
| 6128 |
-
"num_samples":
|
| 6129 |
},
|
| 6130 |
"mean": 0.3798908010369334,
|
| 6131 |
"per_task": {
|
|
@@ -6139,11 +6139,11 @@
|
|
| 6139 |
"metric": "branch_car_at_8",
|
| 6140 |
"micro": {
|
| 6141 |
"confidence": 0.95,
|
| 6142 |
-
"high": 0.
|
| 6143 |
-
"low": 0.
|
| 6144 |
"mean": 0.293583450403069,
|
| 6145 |
"n": 207,
|
| 6146 |
-
"num_samples":
|
| 6147 |
},
|
| 6148 |
"n": 207
|
| 6149 |
},
|
|
@@ -6151,11 +6151,11 @@
|
|
| 6151 |
"macro_by_seed": {
|
| 6152 |
"ci": {
|
| 6153 |
"confidence": 0.95,
|
| 6154 |
-
"high": 0.
|
| 6155 |
-
"low": 0.
|
| 6156 |
"mean": 0.8358156028368796,
|
| 6157 |
"n": 141,
|
| 6158 |
-
"num_samples":
|
| 6159 |
},
|
| 6160 |
"mean": 0.8358156028368796,
|
| 6161 |
"per_seed": {
|
|
@@ -6309,7 +6309,7 @@
|
|
| 6309 |
"low": 0.5941176470588235,
|
| 6310 |
"mean": 0.8647058823529413,
|
| 6311 |
"n": 5,
|
| 6312 |
-
"num_samples":
|
| 6313 |
},
|
| 6314 |
"mean": 0.8647058823529413,
|
| 6315 |
"per_task": {
|
|
@@ -6324,10 +6324,10 @@
|
|
| 6324 |
"micro": {
|
| 6325 |
"confidence": 0.95,
|
| 6326 |
"high": 0.7246376811594203,
|
| 6327 |
-
"low": 0.
|
| 6328 |
"mean": 0.6666666666666666,
|
| 6329 |
"n": 207,
|
| 6330 |
-
"num_samples":
|
| 6331 |
},
|
| 6332 |
"n": 207
|
| 6333 |
},
|
|
@@ -6335,11 +6335,11 @@
|
|
| 6335 |
"macro_by_seed": {
|
| 6336 |
"ci": {
|
| 6337 |
"confidence": 0.95,
|
| 6338 |
-
"high": 1.
|
| 6339 |
-
"low": 1.
|
| 6340 |
"mean": 1.706202811163896,
|
| 6341 |
"n": 141,
|
| 6342 |
-
"num_samples":
|
| 6343 |
},
|
| 6344 |
"mean": 1.706202811163896,
|
| 6345 |
"per_seed": {
|
|
@@ -6493,7 +6493,7 @@
|
|
| 6493 |
"low": 1.2693931020796299,
|
| 6494 |
"mean": 1.7564643673598765,
|
| 6495 |
"n": 5,
|
| 6496 |
-
"num_samples":
|
| 6497 |
},
|
| 6498 |
"mean": 1.7564643673598765,
|
| 6499 |
"per_task": {
|
|
@@ -6507,11 +6507,11 @@
|
|
| 6507 |
"metric": "hidden_chart_oracle_utility_at_8",
|
| 6508 |
"micro": {
|
| 6509 |
"confidence": 0.95,
|
| 6510 |
-
"high": 1.
|
| 6511 |
-
"low": 1.
|
| 6512 |
"mean": 1.3999846732054932,
|
| 6513 |
"n": 207,
|
| 6514 |
-
"num_samples":
|
| 6515 |
},
|
| 6516 |
"n": 207
|
| 6517 |
},
|
|
@@ -6519,11 +6519,11 @@
|
|
| 6519 |
"macro_by_seed": {
|
| 6520 |
"ci": {
|
| 6521 |
"confidence": 0.95,
|
| 6522 |
-
"high": 0.
|
| 6523 |
-
"low": 0.
|
| 6524 |
"mean": 0.4849881796690307,
|
| 6525 |
"n": 141,
|
| 6526 |
-
"num_samples":
|
| 6527 |
},
|
| 6528 |
"mean": 0.4849881796690307,
|
| 6529 |
"per_seed": {
|
|
@@ -6674,10 +6674,10 @@
|
|
| 6674 |
"ci": {
|
| 6675 |
"confidence": 0.95,
|
| 6676 |
"high": 0.6328205128205128,
|
| 6677 |
-
"low": 0.
|
| 6678 |
"mean": 0.5411161387631975,
|
| 6679 |
"n": 5,
|
| 6680 |
-
"num_samples":
|
| 6681 |
},
|
| 6682 |
"mean": 0.5411161387631975,
|
| 6683 |
"per_task": {
|
|
@@ -6691,11 +6691,11 @@
|
|
| 6691 |
"metric": "outcome_ptr_at_8",
|
| 6692 |
"micro": {
|
| 6693 |
"confidence": 0.95,
|
| 6694 |
-
"high": 0.
|
| 6695 |
"low": 0.391304347826087,
|
| 6696 |
"mean": 0.45893719806763283,
|
| 6697 |
"n": 207,
|
| 6698 |
-
"num_samples":
|
| 6699 |
},
|
| 6700 |
"n": 207
|
| 6701 |
},
|
|
@@ -6703,11 +6703,11 @@
|
|
| 6703 |
"macro_by_seed": {
|
| 6704 |
"ci": {
|
| 6705 |
"confidence": 0.95,
|
| 6706 |
-
"high": 0.
|
| 6707 |
-
"low": 0.
|
| 6708 |
"mean": 0.15793930902529071,
|
| 6709 |
"n": 141,
|
| 6710 |
-
"num_samples":
|
| 6711 |
},
|
| 6712 |
"mean": 0.15793930902529071,
|
| 6713 |
"per_seed": {
|
|
@@ -6857,11 +6857,11 @@
|
|
| 6857 |
"macro_by_task": {
|
| 6858 |
"ci": {
|
| 6859 |
"confidence": 0.95,
|
| 6860 |
-
"high": 0.
|
| 6861 |
-
"low": 0.
|
| 6862 |
"mean": 0.16359339232485098,
|
| 6863 |
"n": 5,
|
| 6864 |
-
"num_samples":
|
| 6865 |
},
|
| 6866 |
"mean": 0.16359339232485098,
|
| 6867 |
"per_task": {
|
|
@@ -6875,11 +6875,11 @@
|
|
| 6875 |
"metric": "pairwise_causal_calibration_ece",
|
| 6876 |
"micro": {
|
| 6877 |
"confidence": 0.95,
|
| 6878 |
-
"high": 0.
|
| 6879 |
-
"low": 0.
|
| 6880 |
"mean": 0.16176667684993604,
|
| 6881 |
"n": 207,
|
| 6882 |
-
"num_samples":
|
| 6883 |
},
|
| 6884 |
"n": 207
|
| 6885 |
},
|
|
@@ -6887,11 +6887,11 @@
|
|
| 6887 |
"macro_by_seed": {
|
| 6888 |
"ci": {
|
| 6889 |
"confidence": 0.95,
|
| 6890 |
-
"high": 0.
|
| 6891 |
-
"low": 0.
|
| 6892 |
"mean": 0.47565011820330966,
|
| 6893 |
"n": 141,
|
| 6894 |
-
"num_samples":
|
| 6895 |
},
|
| 6896 |
"mean": 0.47565011820330966,
|
| 6897 |
"per_seed": {
|
|
@@ -7042,10 +7042,10 @@
|
|
| 7042 |
"ci": {
|
| 7043 |
"confidence": 0.95,
|
| 7044 |
"high": 0.5987933634992458,
|
| 7045 |
-
"low": 0.
|
| 7046 |
"mean": 0.3799899446958271,
|
| 7047 |
"n": 5,
|
| 7048 |
-
"num_samples":
|
| 7049 |
},
|
| 7050 |
"mean": 0.3799899446958271,
|
| 7051 |
"per_task": {
|
|
@@ -7059,11 +7059,11 @@
|
|
| 7059 |
"metric": "proposal_oracle_success_at_8",
|
| 7060 |
"micro": {
|
| 7061 |
"confidence": 0.95,
|
| 7062 |
-
"high": 0.
|
| 7063 |
"low": 0.30917874396135264,
|
| 7064 |
"mean": 0.37681159420289856,
|
| 7065 |
"n": 207,
|
| 7066 |
-
"num_samples":
|
| 7067 |
},
|
| 7068 |
"n": 207
|
| 7069 |
},
|
|
@@ -7071,11 +7071,11 @@
|
|
| 7071 |
"macro_by_seed": {
|
| 7072 |
"ci": {
|
| 7073 |
"confidence": 0.95,
|
| 7074 |
-
"high": 0.
|
| 7075 |
-
"low": 0.
|
| 7076 |
"mean": 0.15295508274231676,
|
| 7077 |
"n": 141,
|
| 7078 |
-
"num_samples":
|
| 7079 |
},
|
| 7080 |
"mean": 0.15295508274231676,
|
| 7081 |
"per_seed": {
|
|
@@ -7229,7 +7229,7 @@
|
|
| 7229 |
"low": -0.01581699346405229,
|
| 7230 |
"mean": 0.1427048768225239,
|
| 7231 |
"n": 5,
|
| 7232 |
-
"num_samples":
|
| 7233 |
},
|
| 7234 |
"mean": 0.1427048768225239,
|
| 7235 |
"per_task": {
|
|
@@ -7244,10 +7244,10 @@
|
|
| 7244 |
"micro": {
|
| 7245 |
"confidence": 0.95,
|
| 7246 |
"high": 0.15942028985507245,
|
| 7247 |
-
"low": 0.
|
| 7248 |
"mean": 0.09178743961352658,
|
| 7249 |
"n": 207,
|
| 7250 |
-
"num_samples":
|
| 7251 |
},
|
| 7252 |
"n": 207
|
| 7253 |
},
|
|
@@ -7255,11 +7255,11 @@
|
|
| 7255 |
"macro_by_seed": {
|
| 7256 |
"ci": {
|
| 7257 |
"confidence": 0.95,
|
| 7258 |
-
"high": 1.
|
| 7259 |
-
"low": 0.
|
| 7260 |
"mean": 1.1034544456804372,
|
| 7261 |
"n": 141,
|
| 7262 |
-
"num_samples":
|
| 7263 |
},
|
| 7264 |
"mean": 1.1034544456804372,
|
| 7265 |
"per_seed": {
|
|
@@ -7410,10 +7410,10 @@
|
|
| 7410 |
"ci": {
|
| 7411 |
"confidence": 0.95,
|
| 7412 |
"high": 1.3557187332811518,
|
| 7413 |
-
"low": 0.
|
| 7414 |
"mean": 0.971318265720506,
|
| 7415 |
"n": 5,
|
| 7416 |
-
"num_samples":
|
| 7417 |
},
|
| 7418 |
"mean": 0.971318265720506,
|
| 7419 |
"per_task": {
|
|
@@ -7427,11 +7427,11 @@
|
|
| 7427 |
"metric": "proposal_oracle_utility_at_8",
|
| 7428 |
"micro": {
|
| 7429 |
"confidence": 0.95,
|
| 7430 |
-
"high": 1.
|
| 7431 |
-
"low": 0.
|
| 7432 |
"mean": 0.8810875975700648,
|
| 7433 |
"n": 207,
|
| 7434 |
-
"num_samples":
|
| 7435 |
},
|
| 7436 |
"n": 207
|
| 7437 |
},
|
|
@@ -7440,10 +7440,10 @@
|
|
| 7440 |
"ci": {
|
| 7441 |
"confidence": 0.95,
|
| 7442 |
"high": 0.40175874303166603,
|
| 7443 |
-
"low": 0.
|
| 7444 |
"mean": 0.2610664251762173,
|
| 7445 |
"n": 141,
|
| 7446 |
-
"num_samples":
|
| 7447 |
},
|
| 7448 |
"mean": 0.2610664251762173,
|
| 7449 |
"per_seed": {
|
|
@@ -7593,11 +7593,11 @@
|
|
| 7593 |
"macro_by_task": {
|
| 7594 |
"ci": {
|
| 7595 |
"confidence": 0.95,
|
| 7596 |
-
"high": 0.
|
| 7597 |
-
"low": -0.
|
| 7598 |
"mean": 0.28212140274398645,
|
| 7599 |
"n": 5,
|
| 7600 |
-
"num_samples":
|
| 7601 |
},
|
| 7602 |
"mean": 0.28212140274398645,
|
| 7603 |
"per_task": {
|
|
@@ -7611,11 +7611,11 @@
|
|
| 7611 |
"metric": "proposal_oracle_utility_gain_over_base_at_8",
|
| 7612 |
"micro": {
|
| 7613 |
"confidence": 0.95,
|
| 7614 |
-
"high": 0.
|
| 7615 |
-
"low": 0.
|
| 7616 |
"mean": 0.14528417887851136,
|
| 7617 |
"n": 207,
|
| 7618 |
-
"num_samples":
|
| 7619 |
},
|
| 7620 |
"n": 207
|
| 7621 |
},
|
|
@@ -7623,11 +7623,11 @@
|
|
| 7623 |
"macro_by_seed": {
|
| 7624 |
"ci": {
|
| 7625 |
"confidence": 0.95,
|
| 7626 |
-
"high": 0.
|
| 7627 |
-
"low": 0.
|
| 7628 |
"mean": 0.284160756501182,
|
| 7629 |
"n": 141,
|
| 7630 |
-
"num_samples":
|
| 7631 |
},
|
| 7632 |
"mean": 0.284160756501182,
|
| 7633 |
"per_seed": {
|
|
@@ -7777,11 +7777,11 @@
|
|
| 7777 |
"macro_by_task": {
|
| 7778 |
"ci": {
|
| 7779 |
"confidence": 0.95,
|
| 7780 |
-
"high": 0.
|
| 7781 |
"low": 0.07777777777777777,
|
| 7782 |
"mean": 0.22102564102564098,
|
| 7783 |
"n": 5,
|
| 7784 |
-
"num_samples":
|
| 7785 |
},
|
| 7786 |
"mean": 0.22102564102564098,
|
| 7787 |
"per_task": {
|
|
@@ -7799,7 +7799,7 @@
|
|
| 7799 |
"low": 0.18840579710144928,
|
| 7800 |
"mean": 0.24154589371980675,
|
| 7801 |
"n": 207,
|
| 7802 |
-
"num_samples":
|
| 7803 |
},
|
| 7804 |
"n": 207
|
| 7805 |
},
|
|
@@ -7807,11 +7807,11 @@
|
|
| 7807 |
"macro_by_seed": {
|
| 7808 |
"ci": {
|
| 7809 |
"confidence": 0.95,
|
| 7810 |
-
"high": 0.
|
| 7811 |
-
"low": -0.
|
| 7812 |
"mean": -0.038534278959810875,
|
| 7813 |
"n": 141,
|
| 7814 |
-
"num_samples":
|
| 7815 |
},
|
| 7816 |
"mean": -0.038534278959810875,
|
| 7817 |
"per_seed": {
|
|
@@ -7961,11 +7961,11 @@
|
|
| 7961 |
"macro_by_task": {
|
| 7962 |
"ci": {
|
| 7963 |
"confidence": 0.95,
|
| 7964 |
-
"high": 0.
|
| 7965 |
"low": -0.09986928104575163,
|
| 7966 |
"mean": -0.01625942684766214,
|
| 7967 |
"n": 5,
|
| 7968 |
-
"num_samples":
|
| 7969 |
},
|
| 7970 |
"mean": -0.01625942684766214,
|
| 7971 |
"per_task": {
|
|
@@ -7980,10 +7980,10 @@
|
|
| 7980 |
"micro": {
|
| 7981 |
"confidence": 0.95,
|
| 7982 |
"high": 0.024154589371980676,
|
| 7983 |
-
"low": -0.
|
| 7984 |
"mean": -0.043478260869565216,
|
| 7985 |
"n": 207,
|
| 7986 |
-
"num_samples":
|
| 7987 |
},
|
| 7988 |
"n": 207
|
| 7989 |
},
|
|
@@ -7991,11 +7991,11 @@
|
|
| 7991 |
"macro_by_seed": {
|
| 7992 |
"ci": {
|
| 7993 |
"confidence": 0.95,
|
| 7994 |
-
"high": 0.
|
| 7995 |
-
"low": 0.
|
| 7996 |
"mean": 0.6978370489784622,
|
| 7997 |
"n": 141,
|
| 7998 |
-
"num_samples":
|
| 7999 |
},
|
| 8000 |
"mean": 0.6978370489784622,
|
| 8001 |
"per_seed": {
|
|
@@ -8146,10 +8146,10 @@
|
|
| 8146 |
"ci": {
|
| 8147 |
"confidence": 0.95,
|
| 8148 |
"high": 0.886751612873159,
|
| 8149 |
-
"low": 0.
|
| 8150 |
"mean": 0.5914274646835725,
|
| 8151 |
"n": 5,
|
| 8152 |
-
"num_samples":
|
| 8153 |
},
|
| 8154 |
"mean": 0.5914274646835725,
|
| 8155 |
"per_task": {
|
|
@@ -8163,11 +8163,11 @@
|
|
| 8163 |
"metric": "selected_utility_at_8",
|
| 8164 |
"micro": {
|
| 8165 |
"confidence": 0.95,
|
| 8166 |
-
"high": 0.
|
| 8167 |
-
"low": 0.
|
| 8168 |
"mean": 0.5875041471669958,
|
| 8169 |
"n": 207,
|
| 8170 |
-
"num_samples":
|
| 8171 |
},
|
| 8172 |
"n": 207
|
| 8173 |
},
|
|
@@ -8175,11 +8175,11 @@
|
|
| 8175 |
"macro_by_seed": {
|
| 8176 |
"ci": {
|
| 8177 |
"confidence": 0.95,
|
| 8178 |
-
"high": 0.
|
| 8179 |
-
"low": -0.
|
| 8180 |
"mean": -0.14455097152575763,
|
| 8181 |
"n": 141,
|
| 8182 |
-
"num_samples":
|
| 8183 |
},
|
| 8184 |
"mean": -0.14455097152575763,
|
| 8185 |
"per_seed": {
|
|
@@ -8329,11 +8329,11 @@
|
|
| 8329 |
"macro_by_task": {
|
| 8330 |
"ci": {
|
| 8331 |
"confidence": 0.95,
|
| 8332 |
-
"high": 0.
|
| 8333 |
"low": -0.2553936764363997,
|
| 8334 |
"mean": -0.09776939829294695,
|
| 8335 |
"n": 5,
|
| 8336 |
-
"num_samples":
|
| 8337 |
},
|
| 8338 |
"mean": -0.09776939829294695,
|
| 8339 |
"per_task": {
|
|
@@ -8347,11 +8347,11 @@
|
|
| 8347 |
"metric": "selected_utility_gain_over_base_at_8",
|
| 8348 |
"micro": {
|
| 8349 |
"confidence": 0.95,
|
| 8350 |
-
"high": -0.
|
| 8351 |
"low": -0.26539749033037113,
|
| 8352 |
"mean": -0.14829927152455763,
|
| 8353 |
"n": 207,
|
| 8354 |
-
"num_samples":
|
| 8355 |
},
|
| 8356 |
"n": 207
|
| 8357 |
},
|
|
@@ -8359,11 +8359,11 @@
|
|
| 8359 |
"macro_by_seed": {
|
| 8360 |
"ci": {
|
| 8361 |
"confidence": 0.95,
|
| 8362 |
-
"high": 0.
|
| 8363 |
-
"low": 0.
|
| 8364 |
"mean": 0.40561739670197494,
|
| 8365 |
"n": 141,
|
| 8366 |
-
"num_samples":
|
| 8367 |
},
|
| 8368 |
"mean": 0.40561739670197494,
|
| 8369 |
"per_seed": {
|
|
@@ -8513,11 +8513,11 @@
|
|
| 8513 |
"macro_by_task": {
|
| 8514 |
"ci": {
|
| 8515 |
"confidence": 0.95,
|
| 8516 |
-
"high": 0.
|
| 8517 |
-
"low": 0.
|
| 8518 |
"mean": 0.3798908010369334,
|
| 8519 |
"n": 5,
|
| 8520 |
-
"num_samples":
|
| 8521 |
},
|
| 8522 |
"mean": 0.3798908010369334,
|
| 8523 |
"per_task": {
|
|
@@ -8531,11 +8531,11 @@
|
|
| 8531 |
"metric": "selector_regret_at_8",
|
| 8532 |
"micro": {
|
| 8533 |
"confidence": 0.95,
|
| 8534 |
-
"high": 0.
|
| 8535 |
-
"low": 0.
|
| 8536 |
"mean": 0.293583450403069,
|
| 8537 |
"n": 207,
|
| 8538 |
-
"num_samples":
|
| 8539 |
},
|
| 8540 |
"n": 207
|
| 8541 |
},
|
|
@@ -8543,11 +8543,11 @@
|
|
| 8543 |
"macro_by_seed": {
|
| 8544 |
"ci": {
|
| 8545 |
"confidence": 0.95,
|
| 8546 |
-
"high": 0.
|
| 8547 |
"low": 0.1276595744680851,
|
| 8548 |
"mean": 0.19148936170212766,
|
| 8549 |
"n": 141,
|
| 8550 |
-
"num_samples":
|
| 8551 |
},
|
| 8552 |
"mean": 0.19148936170212766,
|
| 8553 |
"per_seed": {
|
|
@@ -8697,11 +8697,11 @@
|
|
| 8697 |
"macro_by_task": {
|
| 8698 |
"ci": {
|
| 8699 |
"confidence": 0.95,
|
| 8700 |
-
"high": 0.
|
| 8701 |
-
"low": 0.
|
| 8702 |
"mean": 0.15896430367018605,
|
| 8703 |
"n": 5,
|
| 8704 |
-
"num_samples":
|
| 8705 |
},
|
| 8706 |
"mean": 0.15896430367018605,
|
| 8707 |
"per_task": {
|
|
@@ -8715,11 +8715,11 @@
|
|
| 8715 |
"metric": "success_selector_gap_at_8",
|
| 8716 |
"micro": {
|
| 8717 |
"confidence": 0.95,
|
| 8718 |
-
"high": 0.
|
| 8719 |
-
"low": 0.
|
| 8720 |
"mean": 0.13526570048309178,
|
| 8721 |
"n": 207,
|
| 8722 |
-
"num_samples":
|
| 8723 |
},
|
| 8724 |
"n": 207
|
| 8725 |
},
|
|
@@ -8727,11 +8727,11 @@
|
|
| 8727 |
"macro_by_seed": {
|
| 8728 |
"ci": {
|
| 8729 |
"confidence": 0.95,
|
| 8730 |
-
"high": 0.
|
| 8731 |
-
"low": 0.
|
| 8732 |
"mean": 0.3619385342789597,
|
| 8733 |
"n": 141,
|
| 8734 |
-
"num_samples":
|
| 8735 |
},
|
| 8736 |
"mean": 0.3619385342789597,
|
| 8737 |
"per_seed": {
|
|
@@ -8885,7 +8885,7 @@
|
|
| 8885 |
"low": 0.24550025138260434,
|
| 8886 |
"mean": 0.48667672197083967,
|
| 8887 |
"n": 5,
|
| 8888 |
-
"num_samples":
|
| 8889 |
},
|
| 8890 |
"mean": 0.48667672197083967,
|
| 8891 |
"per_task": {
|
|
@@ -8899,11 +8899,11 @@
|
|
| 8899 |
"metric": "success_support_gap_at_8",
|
| 8900 |
"micro": {
|
| 8901 |
"confidence": 0.95,
|
| 8902 |
-
"high": 0.
|
| 8903 |
"low": 0.2318840579710145,
|
| 8904 |
"mean": 0.2946859903381642,
|
| 8905 |
"n": 207,
|
| 8906 |
-
"num_samples":
|
| 8907 |
},
|
| 8908 |
"n": 207
|
| 8909 |
},
|
|
@@ -8911,11 +8911,11 @@
|
|
| 8911 |
"macro_by_seed": {
|
| 8912 |
"ci": {
|
| 8913 |
"confidence": 0.95,
|
| 8914 |
-
"high": 0.
|
| 8915 |
-
"low": 0.
|
| 8916 |
"mean": 0.5534278959810874,
|
| 8917 |
"n": 141,
|
| 8918 |
-
"num_samples":
|
| 8919 |
},
|
| 8920 |
"mean": 0.5534278959810874,
|
| 8921 |
"per_seed": {
|
|
@@ -9065,11 +9065,11 @@
|
|
| 9065 |
"macro_by_task": {
|
| 9066 |
"ci": {
|
| 9067 |
"confidence": 0.95,
|
| 9068 |
-
"high": 0.
|
| 9069 |
-
"low": 0.
|
| 9070 |
"mean": 0.6456410256410257,
|
| 9071 |
"n": 5,
|
| 9072 |
-
"num_samples":
|
| 9073 |
},
|
| 9074 |
"mean": 0.6456410256410257,
|
| 9075 |
"per_task": {
|
|
@@ -9087,7 +9087,7 @@
|
|
| 9087 |
"low": 0.357487922705314,
|
| 9088 |
"mean": 0.42995169082125606,
|
| 9089 |
"n": 207,
|
| 9090 |
-
"num_samples":
|
| 9091 |
},
|
| 9092 |
"n": 207
|
| 9093 |
},
|
|
@@ -9095,11 +9095,11 @@
|
|
| 9095 |
"macro_by_seed": {
|
| 9096 |
"ci": {
|
| 9097 |
"confidence": 0.95,
|
| 9098 |
-
"high": 0.
|
| 9099 |
-
"low": 0.
|
| 9100 |
"mean": 0.6062044484887313,
|
| 9101 |
"n": 141,
|
| 9102 |
-
"num_samples":
|
| 9103 |
},
|
| 9104 |
"mean": 0.6062044484887313,
|
| 9105 |
"per_seed": {
|
|
@@ -9253,7 +9253,7 @@
|
|
| 9253 |
"low": 0.4585766474838634,
|
| 9254 |
"mean": 0.788961660718036,
|
| 9255 |
"n": 5,
|
| 9256 |
-
"num_samples":
|
| 9257 |
},
|
| 9258 |
"mean": 0.788961660718036,
|
| 9259 |
"per_task": {
|
|
@@ -9267,11 +9267,11 @@
|
|
| 9267 |
"metric": "support_gap_at_8",
|
| 9268 |
"micro": {
|
| 9269 |
"confidence": 0.95,
|
| 9270 |
-
"high": 0.
|
| 9271 |
"low": 0.4234259423213799,
|
| 9272 |
"mean": 0.5282977284379371,
|
| 9273 |
"n": 207,
|
| 9274 |
-
"num_samples":
|
| 9275 |
},
|
| 9276 |
"n": 207
|
| 9277 |
},
|
|
@@ -9279,11 +9279,11 @@
|
|
| 9279 |
"macro_by_seed": {
|
| 9280 |
"ci": {
|
| 9281 |
"confidence": 0.95,
|
| 9282 |
-
"high": 1.
|
| 9283 |
-
"low": 0.
|
| 9284 |
"mean": 1.0115917835033712,
|
| 9285 |
"n": 141,
|
| 9286 |
-
"num_samples":
|
| 9287 |
},
|
| 9288 |
"mean": 1.0115917835033712,
|
| 9289 |
"per_seed": {
|
|
@@ -9437,7 +9437,7 @@
|
|
| 9437 |
"low": 0.722751199278172,
|
| 9438 |
"mean": 1.1686045027220224,
|
| 9439 |
"n": 5,
|
| 9440 |
-
"num_samples":
|
| 9441 |
},
|
| 9442 |
"mean": 1.1686045027220224,
|
| 9443 |
"per_task": {
|
|
@@ -9451,11 +9451,11 @@
|
|
| 9451 |
"metric": "total_car_to_hidden_at_8",
|
| 9452 |
"micro": {
|
| 9453 |
"confidence": 0.95,
|
| 9454 |
-
"high": 0.
|
| 9455 |
-
"low": 0.
|
| 9456 |
"mean": 0.8212702652815715,
|
| 9457 |
"n": 207,
|
| 9458 |
-
"num_samples":
|
| 9459 |
},
|
| 9460 |
"n": 207
|
| 9461 |
}
|
|
|
|
| 5599 |
"macro_by_seed": {
|
| 5600 |
"ci": {
|
| 5601 |
"confidence": 0.95,
|
| 5602 |
+
"high": 0.3918439716312056,
|
| 5603 |
+
"low": 0.25,
|
| 5604 |
"mean": 0.322695035460993,
|
| 5605 |
"n": 141,
|
| 5606 |
+
"num_samples": 1000
|
| 5607 |
},
|
| 5608 |
"mean": 0.322695035460993,
|
| 5609 |
"per_seed": {
|
|
|
|
| 5757 |
"low": 0.10666666666666666,
|
| 5758 |
"mean": 0.23728506787330317,
|
| 5759 |
"n": 5,
|
| 5760 |
+
"num_samples": 1000
|
| 5761 |
},
|
| 5762 |
"mean": 0.23728506787330317,
|
| 5763 |
"per_task": {
|
|
|
|
| 5775 |
"low": 0.22705314009661837,
|
| 5776 |
"mean": 0.28502415458937197,
|
| 5777 |
"n": 207,
|
| 5778 |
+
"num_samples": 1000
|
| 5779 |
},
|
| 5780 |
"n": 207
|
| 5781 |
},
|
|
|
|
| 5783 |
"macro_by_seed": {
|
| 5784 |
"ci": {
|
| 5785 |
"confidence": 0.95,
|
| 5786 |
+
"high": 0.9671404774753207,
|
| 5787 |
+
"low": 0.7101003489727332,
|
| 5788 |
"mean": 0.84238802050422,
|
| 5789 |
"n": 141,
|
| 5790 |
+
"num_samples": 1000
|
| 5791 |
},
|
| 5792 |
"mean": 0.84238802050422,
|
| 5793 |
"per_seed": {
|
|
|
|
| 5937 |
"macro_by_task": {
|
| 5938 |
"ci": {
|
| 5939 |
"confidence": 0.95,
|
| 5940 |
+
"high": 0.924619877206718,
|
| 5941 |
+
"low": 0.40042830170729216,
|
| 5942 |
"mean": 0.6891968629765195,
|
| 5943 |
"n": 5,
|
| 5944 |
+
"num_samples": 1000
|
| 5945 |
},
|
| 5946 |
"mean": 0.6891968629765195,
|
| 5947 |
"per_task": {
|
|
|
|
| 5956 |
"micro": {
|
| 5957 |
"confidence": 0.95,
|
| 5958 |
"high": 0.851249745030154,
|
| 5959 |
+
"low": 0.625559344417573,
|
| 5960 |
"mean": 0.7358034186915535,
|
| 5961 |
"n": 207,
|
| 5962 |
+
"num_samples": 1000
|
| 5963 |
},
|
| 5964 |
"n": 207
|
| 5965 |
},
|
|
|
|
| 5967 |
"macro_by_seed": {
|
| 5968 |
"ci": {
|
| 5969 |
"confidence": 0.95,
|
| 5970 |
+
"high": 0.5194647902708409,
|
| 5971 |
+
"low": 0.2999603762294949,
|
| 5972 |
"mean": 0.40561739670197494,
|
| 5973 |
"n": 141,
|
| 5974 |
+
"num_samples": 1000
|
| 5975 |
},
|
| 5976 |
"mean": 0.40561739670197494,
|
| 5977 |
"per_seed": {
|
|
|
|
| 6121 |
"macro_by_task": {
|
| 6122 |
"ci": {
|
| 6123 |
"confidence": 0.95,
|
| 6124 |
+
"high": 0.5319262530406317,
|
| 6125 |
+
"low": 0.19367229081819662,
|
| 6126 |
"mean": 0.3798908010369334,
|
| 6127 |
"n": 5,
|
| 6128 |
+
"num_samples": 1000
|
| 6129 |
},
|
| 6130 |
"mean": 0.3798908010369334,
|
| 6131 |
"per_task": {
|
|
|
|
| 6139 |
"metric": "branch_car_at_8",
|
| 6140 |
"micro": {
|
| 6141 |
"confidence": 0.95,
|
| 6142 |
+
"high": 0.3721372773162211,
|
| 6143 |
+
"low": 0.21313571063115527,
|
| 6144 |
"mean": 0.293583450403069,
|
| 6145 |
"n": 207,
|
| 6146 |
+
"num_samples": 1000
|
| 6147 |
},
|
| 6148 |
"n": 207
|
| 6149 |
},
|
|
|
|
| 6151 |
"macro_by_seed": {
|
| 6152 |
"ci": {
|
| 6153 |
"confidence": 0.95,
|
| 6154 |
+
"high": 0.8887706855791963,
|
| 6155 |
+
"low": 0.7754137115839246,
|
| 6156 |
"mean": 0.8358156028368796,
|
| 6157 |
"n": 141,
|
| 6158 |
+
"num_samples": 1000
|
| 6159 |
},
|
| 6160 |
"mean": 0.8358156028368796,
|
| 6161 |
"per_seed": {
|
|
|
|
| 6309 |
"low": 0.5941176470588235,
|
| 6310 |
"mean": 0.8647058823529413,
|
| 6311 |
"n": 5,
|
| 6312 |
+
"num_samples": 1000
|
| 6313 |
},
|
| 6314 |
"mean": 0.8647058823529413,
|
| 6315 |
"per_task": {
|
|
|
|
| 6324 |
"micro": {
|
| 6325 |
"confidence": 0.95,
|
| 6326 |
"high": 0.7246376811594203,
|
| 6327 |
+
"low": 0.5990338164251208,
|
| 6328 |
"mean": 0.6666666666666666,
|
| 6329 |
"n": 207,
|
| 6330 |
+
"num_samples": 1000
|
| 6331 |
},
|
| 6332 |
"n": 207
|
| 6333 |
},
|
|
|
|
| 6335 |
"macro_by_seed": {
|
| 6336 |
"ci": {
|
| 6337 |
"confidence": 0.95,
|
| 6338 |
+
"high": 1.8010394367889166,
|
| 6339 |
+
"low": 1.5995275395488624,
|
| 6340 |
"mean": 1.706202811163896,
|
| 6341 |
"n": 141,
|
| 6342 |
+
"num_samples": 1000
|
| 6343 |
},
|
| 6344 |
"mean": 1.706202811163896,
|
| 6345 |
"per_seed": {
|
|
|
|
| 6493 |
"low": 1.2693931020796299,
|
| 6494 |
"mean": 1.7564643673598765,
|
| 6495 |
"n": 5,
|
| 6496 |
+
"num_samples": 1000
|
| 6497 |
},
|
| 6498 |
"mean": 1.7564643673598765,
|
| 6499 |
"per_task": {
|
|
|
|
| 6507 |
"metric": "hidden_chart_oracle_utility_at_8",
|
| 6508 |
"micro": {
|
| 6509 |
"confidence": 0.95,
|
| 6510 |
+
"high": 1.5062634588583657,
|
| 6511 |
+
"low": 1.2799855626507657,
|
| 6512 |
"mean": 1.3999846732054932,
|
| 6513 |
"n": 207,
|
| 6514 |
+
"num_samples": 1000
|
| 6515 |
},
|
| 6516 |
"n": 207
|
| 6517 |
},
|
|
|
|
| 6519 |
"macro_by_seed": {
|
| 6520 |
"ci": {
|
| 6521 |
"confidence": 0.95,
|
| 6522 |
+
"high": 0.5674940898345152,
|
| 6523 |
+
"low": 0.4113475177304965,
|
| 6524 |
"mean": 0.4849881796690307,
|
| 6525 |
"n": 141,
|
| 6526 |
+
"num_samples": 1000
|
| 6527 |
},
|
| 6528 |
"mean": 0.4849881796690307,
|
| 6529 |
"per_seed": {
|
|
|
|
| 6674 |
"ci": {
|
| 6675 |
"confidence": 0.95,
|
| 6676 |
"high": 0.6328205128205128,
|
| 6677 |
+
"low": 0.4294117647058823,
|
| 6678 |
"mean": 0.5411161387631975,
|
| 6679 |
"n": 5,
|
| 6680 |
+
"num_samples": 1000
|
| 6681 |
},
|
| 6682 |
"mean": 0.5411161387631975,
|
| 6683 |
"per_task": {
|
|
|
|
| 6691 |
"metric": "outcome_ptr_at_8",
|
| 6692 |
"micro": {
|
| 6693 |
"confidence": 0.95,
|
| 6694 |
+
"high": 0.5265700483091788,
|
| 6695 |
"low": 0.391304347826087,
|
| 6696 |
"mean": 0.45893719806763283,
|
| 6697 |
"n": 207,
|
| 6698 |
+
"num_samples": 1000
|
| 6699 |
},
|
| 6700 |
"n": 207
|
| 6701 |
},
|
|
|
|
| 6703 |
"macro_by_seed": {
|
| 6704 |
"ci": {
|
| 6705 |
"confidence": 0.95,
|
| 6706 |
+
"high": 0.17446188421793687,
|
| 6707 |
+
"low": 0.14114248449490324,
|
| 6708 |
"mean": 0.15793930902529071,
|
| 6709 |
"n": 141,
|
| 6710 |
+
"num_samples": 1000
|
| 6711 |
},
|
| 6712 |
"mean": 0.15793930902529071,
|
| 6713 |
"per_seed": {
|
|
|
|
| 6857 |
"macro_by_task": {
|
| 6858 |
"ci": {
|
| 6859 |
"confidence": 0.95,
|
| 6860 |
+
"high": 0.17042242389065887,
|
| 6861 |
+
"low": 0.1574042611063619,
|
| 6862 |
"mean": 0.16359339232485098,
|
| 6863 |
"n": 5,
|
| 6864 |
+
"num_samples": 1000
|
| 6865 |
},
|
| 6866 |
"mean": 0.16359339232485098,
|
| 6867 |
"per_task": {
|
|
|
|
| 6875 |
"metric": "pairwise_causal_calibration_ece",
|
| 6876 |
"micro": {
|
| 6877 |
"confidence": 0.95,
|
| 6878 |
+
"high": 0.1771431506285629,
|
| 6879 |
+
"low": 0.14633605302901057,
|
| 6880 |
"mean": 0.16176667684993604,
|
| 6881 |
"n": 207,
|
| 6882 |
+
"num_samples": 1000
|
| 6883 |
},
|
| 6884 |
"n": 207
|
| 6885 |
},
|
|
|
|
| 6887 |
"macro_by_seed": {
|
| 6888 |
"ci": {
|
| 6889 |
"confidence": 0.95,
|
| 6890 |
+
"high": 0.5508274231678488,
|
| 6891 |
+
"low": 0.3905437352245863,
|
| 6892 |
"mean": 0.47565011820330966,
|
| 6893 |
"n": 141,
|
| 6894 |
+
"num_samples": 1000
|
| 6895 |
},
|
| 6896 |
"mean": 0.47565011820330966,
|
| 6897 |
"per_seed": {
|
|
|
|
| 7042 |
"ci": {
|
| 7043 |
"confidence": 0.95,
|
| 7044 |
"high": 0.5987933634992458,
|
| 7045 |
+
"low": 0.13725490196078433,
|
| 7046 |
"mean": 0.3799899446958271,
|
| 7047 |
"n": 5,
|
| 7048 |
+
"num_samples": 1000
|
| 7049 |
},
|
| 7050 |
"mean": 0.3799899446958271,
|
| 7051 |
"per_task": {
|
|
|
|
| 7059 |
"metric": "proposal_oracle_success_at_8",
|
| 7060 |
"micro": {
|
| 7061 |
"confidence": 0.95,
|
| 7062 |
+
"high": 0.4396135265700483,
|
| 7063 |
"low": 0.30917874396135264,
|
| 7064 |
"mean": 0.37681159420289856,
|
| 7065 |
"n": 207,
|
| 7066 |
+
"num_samples": 1000
|
| 7067 |
},
|
| 7068 |
"n": 207
|
| 7069 |
},
|
|
|
|
| 7071 |
"macro_by_seed": {
|
| 7072 |
"ci": {
|
| 7073 |
"confidence": 0.95,
|
| 7074 |
+
"high": 0.23179669030732858,
|
| 7075 |
+
"low": 0.06643026004728132,
|
| 7076 |
"mean": 0.15295508274231676,
|
| 7077 |
"n": 141,
|
| 7078 |
+
"num_samples": 1000
|
| 7079 |
},
|
| 7080 |
"mean": 0.15295508274231676,
|
| 7081 |
"per_seed": {
|
|
|
|
| 7229 |
"low": -0.01581699346405229,
|
| 7230 |
"mean": 0.1427048768225239,
|
| 7231 |
"n": 5,
|
| 7232 |
+
"num_samples": 1000
|
| 7233 |
},
|
| 7234 |
"mean": 0.1427048768225239,
|
| 7235 |
"per_task": {
|
|
|
|
| 7244 |
"micro": {
|
| 7245 |
"confidence": 0.95,
|
| 7246 |
"high": 0.15942028985507245,
|
| 7247 |
+
"low": 0.024154589371980676,
|
| 7248 |
"mean": 0.09178743961352658,
|
| 7249 |
"n": 207,
|
| 7250 |
+
"num_samples": 1000
|
| 7251 |
},
|
| 7252 |
"n": 207
|
| 7253 |
},
|
|
|
|
| 7255 |
"macro_by_seed": {
|
| 7256 |
"ci": {
|
| 7257 |
"confidence": 0.95,
|
| 7258 |
+
"high": 1.2371345245170386,
|
| 7259 |
+
"low": 0.9570726445954775,
|
| 7260 |
"mean": 1.1034544456804372,
|
| 7261 |
"n": 141,
|
| 7262 |
+
"num_samples": 1000
|
| 7263 |
},
|
| 7264 |
"mean": 1.1034544456804372,
|
| 7265 |
"per_seed": {
|
|
|
|
| 7410 |
"ci": {
|
| 7411 |
"confidence": 0.95,
|
| 7412 |
"high": 1.3557187332811518,
|
| 7413 |
+
"low": 0.5600978099637561,
|
| 7414 |
"mean": 0.971318265720506,
|
| 7415 |
"n": 5,
|
| 7416 |
+
"num_samples": 1000
|
| 7417 |
},
|
| 7418 |
"mean": 0.971318265720506,
|
| 7419 |
"per_task": {
|
|
|
|
| 7427 |
"metric": "proposal_oracle_utility_at_8",
|
| 7428 |
"micro": {
|
| 7429 |
"confidence": 0.95,
|
| 7430 |
+
"high": 1.0026757181962336,
|
| 7431 |
+
"low": 0.7613600684022801,
|
| 7432 |
"mean": 0.8810875975700648,
|
| 7433 |
"n": 207,
|
| 7434 |
+
"num_samples": 1000
|
| 7435 |
},
|
| 7436 |
"n": 207
|
| 7437 |
},
|
|
|
|
| 7440 |
"ci": {
|
| 7441 |
"confidence": 0.95,
|
| 7442 |
"high": 0.40175874303166603,
|
| 7443 |
+
"low": 0.11287717537031931,
|
| 7444 |
"mean": 0.2610664251762173,
|
| 7445 |
"n": 141,
|
| 7446 |
+
"num_samples": 1000
|
| 7447 |
},
|
| 7448 |
"mean": 0.2610664251762173,
|
| 7449 |
"per_seed": {
|
|
|
|
| 7593 |
"macro_by_task": {
|
| 7594 |
"ci": {
|
| 7595 |
"confidence": 0.95,
|
| 7596 |
+
"high": 0.5660580151865625,
|
| 7597 |
+
"low": -0.016164922901372936,
|
| 7598 |
"mean": 0.28212140274398645,
|
| 7599 |
"n": 5,
|
| 7600 |
+
"num_samples": 1000
|
| 7601 |
},
|
| 7602 |
"mean": 0.28212140274398645,
|
| 7603 |
"per_task": {
|
|
|
|
| 7611 |
"metric": "proposal_oracle_utility_gain_over_base_at_8",
|
| 7612 |
"micro": {
|
| 7613 |
"confidence": 0.95,
|
| 7614 |
+
"high": 0.2609384466434496,
|
| 7615 |
+
"low": 0.03087088158933428,
|
| 7616 |
"mean": 0.14528417887851136,
|
| 7617 |
"n": 207,
|
| 7618 |
+
"num_samples": 1000
|
| 7619 |
},
|
| 7620 |
"n": 207
|
| 7621 |
},
|
|
|
|
| 7623 |
"macro_by_seed": {
|
| 7624 |
"ci": {
|
| 7625 |
"confidence": 0.95,
|
| 7626 |
+
"high": 0.35579196217494086,
|
| 7627 |
+
"low": 0.2118203309692671,
|
| 7628 |
"mean": 0.284160756501182,
|
| 7629 |
"n": 141,
|
| 7630 |
+
"num_samples": 1000
|
| 7631 |
},
|
| 7632 |
"mean": 0.284160756501182,
|
| 7633 |
"per_seed": {
|
|
|
|
| 7777 |
"macro_by_task": {
|
| 7778 |
"ci": {
|
| 7779 |
"confidence": 0.95,
|
| 7780 |
+
"high": 0.391965811965812,
|
| 7781 |
"low": 0.07777777777777777,
|
| 7782 |
"mean": 0.22102564102564098,
|
| 7783 |
"n": 5,
|
| 7784 |
+
"num_samples": 1000
|
| 7785 |
},
|
| 7786 |
"mean": 0.22102564102564098,
|
| 7787 |
"per_task": {
|
|
|
|
| 7799 |
"low": 0.18840579710144928,
|
| 7800 |
"mean": 0.24154589371980675,
|
| 7801 |
"n": 207,
|
| 7802 |
+
"num_samples": 1000
|
| 7803 |
},
|
| 7804 |
"n": 207
|
| 7805 |
},
|
|
|
|
| 7807 |
"macro_by_seed": {
|
| 7808 |
"ci": {
|
| 7809 |
"confidence": 0.95,
|
| 7810 |
+
"high": 0.04893617021276596,
|
| 7811 |
+
"low": -0.12411347517730496,
|
| 7812 |
"mean": -0.038534278959810875,
|
| 7813 |
"n": 141,
|
| 7814 |
+
"num_samples": 1000
|
| 7815 |
},
|
| 7816 |
"mean": -0.038534278959810875,
|
| 7817 |
"per_seed": {
|
|
|
|
| 7961 |
"macro_by_task": {
|
| 7962 |
"ci": {
|
| 7963 |
"confidence": 0.95,
|
| 7964 |
+
"high": 0.08220211161387632,
|
| 7965 |
"low": -0.09986928104575163,
|
| 7966 |
"mean": -0.01625942684766214,
|
| 7967 |
"n": 5,
|
| 7968 |
+
"num_samples": 1000
|
| 7969 |
},
|
| 7970 |
"mean": -0.01625942684766214,
|
| 7971 |
"per_task": {
|
|
|
|
| 7980 |
"micro": {
|
| 7981 |
"confidence": 0.95,
|
| 7982 |
"high": 0.024154589371980676,
|
| 7983 |
+
"low": -0.10628019323671498,
|
| 7984 |
"mean": -0.043478260869565216,
|
| 7985 |
"n": 207,
|
| 7986 |
+
"num_samples": 1000
|
| 7987 |
},
|
| 7988 |
"n": 207
|
| 7989 |
},
|
|
|
|
| 7991 |
"macro_by_seed": {
|
| 7992 |
"ci": {
|
| 7993 |
"confidence": 0.95,
|
| 7994 |
+
"high": 0.8302852093747126,
|
| 7995 |
+
"low": 0.5613265780814601,
|
| 7996 |
"mean": 0.6978370489784622,
|
| 7997 |
"n": 141,
|
| 7998 |
+
"num_samples": 1000
|
| 7999 |
},
|
| 8000 |
"mean": 0.6978370489784622,
|
| 8001 |
"per_seed": {
|
|
|
|
| 8146 |
"ci": {
|
| 8147 |
"confidence": 0.95,
|
| 8148 |
"high": 0.886751612873159,
|
| 8149 |
+
"low": 0.2938098484889055,
|
| 8150 |
"mean": 0.5914274646835725,
|
| 8151 |
"n": 5,
|
| 8152 |
+
"num_samples": 1000
|
| 8153 |
},
|
| 8154 |
"mean": 0.5914274646835725,
|
| 8155 |
"per_task": {
|
|
|
|
| 8163 |
"metric": "selected_utility_at_8",
|
| 8164 |
"micro": {
|
| 8165 |
"confidence": 0.95,
|
| 8166 |
+
"high": 0.7017283821891844,
|
| 8167 |
+
"low": 0.48462916244437765,
|
| 8168 |
"mean": 0.5875041471669958,
|
| 8169 |
"n": 207,
|
| 8170 |
+
"num_samples": 1000
|
| 8171 |
},
|
| 8172 |
"n": 207
|
| 8173 |
},
|
|
|
|
| 8175 |
"macro_by_seed": {
|
| 8176 |
"ci": {
|
| 8177 |
"confidence": 0.95,
|
| 8178 |
+
"high": 0.01420507583930653,
|
| 8179 |
+
"low": -0.29225106954879526,
|
| 8180 |
"mean": -0.14455097152575763,
|
| 8181 |
"n": 141,
|
| 8182 |
+
"num_samples": 1000
|
| 8183 |
},
|
| 8184 |
"mean": -0.14455097152575763,
|
| 8185 |
"per_seed": {
|
|
|
|
| 8329 |
"macro_by_task": {
|
| 8330 |
"ci": {
|
| 8331 |
"confidence": 0.95,
|
| 8332 |
+
"high": 0.10282889067617237,
|
| 8333 |
"low": -0.2553936764363997,
|
| 8334 |
"mean": -0.09776939829294695,
|
| 8335 |
"n": 5,
|
| 8336 |
+
"num_samples": 1000
|
| 8337 |
},
|
| 8338 |
"mean": -0.09776939829294695,
|
| 8339 |
"per_task": {
|
|
|
|
| 8347 |
"metric": "selected_utility_gain_over_base_at_8",
|
| 8348 |
"micro": {
|
| 8349 |
"confidence": 0.95,
|
| 8350 |
+
"high": -0.023960533527688885,
|
| 8351 |
"low": -0.26539749033037113,
|
| 8352 |
"mean": -0.14829927152455763,
|
| 8353 |
"n": 207,
|
| 8354 |
+
"num_samples": 1000
|
| 8355 |
},
|
| 8356 |
"n": 207
|
| 8357 |
},
|
|
|
|
| 8359 |
"macro_by_seed": {
|
| 8360 |
"ci": {
|
| 8361 |
"confidence": 0.95,
|
| 8362 |
+
"high": 0.5194647902708409,
|
| 8363 |
+
"low": 0.2999603762294949,
|
| 8364 |
"mean": 0.40561739670197494,
|
| 8365 |
"n": 141,
|
| 8366 |
+
"num_samples": 1000
|
| 8367 |
},
|
| 8368 |
"mean": 0.40561739670197494,
|
| 8369 |
"per_seed": {
|
|
|
|
| 8513 |
"macro_by_task": {
|
| 8514 |
"ci": {
|
| 8515 |
"confidence": 0.95,
|
| 8516 |
+
"high": 0.5319262530406317,
|
| 8517 |
+
"low": 0.19367229081819662,
|
| 8518 |
"mean": 0.3798908010369334,
|
| 8519 |
"n": 5,
|
| 8520 |
+
"num_samples": 1000
|
| 8521 |
},
|
| 8522 |
"mean": 0.3798908010369334,
|
| 8523 |
"per_task": {
|
|
|
|
| 8531 |
"metric": "selector_regret_at_8",
|
| 8532 |
"micro": {
|
| 8533 |
"confidence": 0.95,
|
| 8534 |
+
"high": 0.3721372773162211,
|
| 8535 |
+
"low": 0.21313571063115527,
|
| 8536 |
"mean": 0.293583450403069,
|
| 8537 |
"n": 207,
|
| 8538 |
+
"num_samples": 1000
|
| 8539 |
},
|
| 8540 |
"n": 207
|
| 8541 |
},
|
|
|
|
| 8543 |
"macro_by_seed": {
|
| 8544 |
"ci": {
|
| 8545 |
"confidence": 0.95,
|
| 8546 |
+
"high": 0.2553191489361702,
|
| 8547 |
"low": 0.1276595744680851,
|
| 8548 |
"mean": 0.19148936170212766,
|
| 8549 |
"n": 141,
|
| 8550 |
+
"num_samples": 1000
|
| 8551 |
},
|
| 8552 |
"mean": 0.19148936170212766,
|
| 8553 |
"per_seed": {
|
|
|
|
| 8697 |
"macro_by_task": {
|
| 8698 |
"ci": {
|
| 8699 |
"confidence": 0.95,
|
| 8700 |
+
"high": 0.26170940170940166,
|
| 8701 |
+
"low": 0.05399698340874811,
|
| 8702 |
"mean": 0.15896430367018605,
|
| 8703 |
"n": 5,
|
| 8704 |
+
"num_samples": 1000
|
| 8705 |
},
|
| 8706 |
"mean": 0.15896430367018605,
|
| 8707 |
"per_task": {
|
|
|
|
| 8715 |
"metric": "success_selector_gap_at_8",
|
| 8716 |
"micro": {
|
| 8717 |
"confidence": 0.95,
|
| 8718 |
+
"high": 0.18357487922705315,
|
| 8719 |
+
"low": 0.09178743961352658,
|
| 8720 |
"mean": 0.13526570048309178,
|
| 8721 |
"n": 207,
|
| 8722 |
+
"num_samples": 1000
|
| 8723 |
},
|
| 8724 |
"n": 207
|
| 8725 |
},
|
|
|
|
| 8727 |
"macro_by_seed": {
|
| 8728 |
"ci": {
|
| 8729 |
"confidence": 0.95,
|
| 8730 |
+
"high": 0.43794326241134757,
|
| 8731 |
+
"low": 0.29089834515366436,
|
| 8732 |
"mean": 0.3619385342789597,
|
| 8733 |
"n": 141,
|
| 8734 |
+
"num_samples": 1000
|
| 8735 |
},
|
| 8736 |
"mean": 0.3619385342789597,
|
| 8737 |
"per_seed": {
|
|
|
|
| 8885 |
"low": 0.24550025138260434,
|
| 8886 |
"mean": 0.48667672197083967,
|
| 8887 |
"n": 5,
|
| 8888 |
+
"num_samples": 1000
|
| 8889 |
},
|
| 8890 |
"mean": 0.48667672197083967,
|
| 8891 |
"per_task": {
|
|
|
|
| 8899 |
"metric": "success_support_gap_at_8",
|
| 8900 |
"micro": {
|
| 8901 |
"confidence": 0.95,
|
| 8902 |
+
"high": 0.357487922705314,
|
| 8903 |
"low": 0.2318840579710145,
|
| 8904 |
"mean": 0.2946859903381642,
|
| 8905 |
"n": 207,
|
| 8906 |
+
"num_samples": 1000
|
| 8907 |
},
|
| 8908 |
"n": 207
|
| 8909 |
},
|
|
|
|
| 8911 |
"macro_by_seed": {
|
| 8912 |
"ci": {
|
| 8913 |
"confidence": 0.95,
|
| 8914 |
+
"high": 0.6296690307328605,
|
| 8915 |
+
"low": 0.47482269503546104,
|
| 8916 |
"mean": 0.5534278959810874,
|
| 8917 |
"n": 141,
|
| 8918 |
+
"num_samples": 1000
|
| 8919 |
},
|
| 8920 |
"mean": 0.5534278959810874,
|
| 8921 |
"per_seed": {
|
|
|
|
| 9065 |
"macro_by_task": {
|
| 9066 |
"ci": {
|
| 9067 |
"confidence": 0.95,
|
| 9068 |
+
"high": 0.8844444444444445,
|
| 9069 |
+
"low": 0.3567521367521368,
|
| 9070 |
"mean": 0.6456410256410257,
|
| 9071 |
"n": 5,
|
| 9072 |
+
"num_samples": 1000
|
| 9073 |
},
|
| 9074 |
"mean": 0.6456410256410257,
|
| 9075 |
"per_task": {
|
|
|
|
| 9087 |
"low": 0.357487922705314,
|
| 9088 |
"mean": 0.42995169082125606,
|
| 9089 |
"n": 207,
|
| 9090 |
+
"num_samples": 1000
|
| 9091 |
},
|
| 9092 |
"n": 207
|
| 9093 |
},
|
|
|
|
| 9095 |
"macro_by_seed": {
|
| 9096 |
"ci": {
|
| 9097 |
"confidence": 0.95,
|
| 9098 |
+
"high": 0.7276497844564805,
|
| 9099 |
+
"low": 0.49026043312640605,
|
| 9100 |
"mean": 0.6062044484887313,
|
| 9101 |
"n": 141,
|
| 9102 |
+
"num_samples": 1000
|
| 9103 |
},
|
| 9104 |
"mean": 0.6062044484887313,
|
| 9105 |
"per_seed": {
|
|
|
|
| 9253 |
"low": 0.4585766474838634,
|
| 9254 |
"mean": 0.788961660718036,
|
| 9255 |
"n": 5,
|
| 9256 |
+
"num_samples": 1000
|
| 9257 |
},
|
| 9258 |
"mean": 0.788961660718036,
|
| 9259 |
"per_task": {
|
|
|
|
| 9267 |
"metric": "support_gap_at_8",
|
| 9268 |
"micro": {
|
| 9269 |
"confidence": 0.95,
|
| 9270 |
+
"high": 0.6340587634713039,
|
| 9271 |
"low": 0.4234259423213799,
|
| 9272 |
"mean": 0.5282977284379371,
|
| 9273 |
"n": 207,
|
| 9274 |
+
"num_samples": 1000
|
| 9275 |
},
|
| 9276 |
"n": 207
|
| 9277 |
},
|
|
|
|
| 9279 |
"macro_by_seed": {
|
| 9280 |
"ci": {
|
| 9281 |
"confidence": 0.95,
|
| 9282 |
+
"high": 1.141025779891047,
|
| 9283 |
+
"low": 0.8745643807700053,
|
| 9284 |
"mean": 1.0115917835033712,
|
| 9285 |
"n": 141,
|
| 9286 |
+
"num_samples": 1000
|
| 9287 |
},
|
| 9288 |
"mean": 1.0115917835033712,
|
| 9289 |
"per_seed": {
|
|
|
|
| 9437 |
"low": 0.722751199278172,
|
| 9438 |
"mean": 1.1686045027220224,
|
| 9439 |
"n": 5,
|
| 9440 |
+
"num_samples": 1000
|
| 9441 |
},
|
| 9442 |
"mean": 1.1686045027220224,
|
| 9443 |
"per_task": {
|
|
|
|
| 9451 |
"metric": "total_car_to_hidden_at_8",
|
| 9452 |
"micro": {
|
| 9453 |
"confidence": 0.95,
|
| 9454 |
+
"high": 0.9380562976314821,
|
| 9455 |
+
"low": 0.6983931870469688,
|
| 9456 |
"mean": 0.8212702652815715,
|
| 9457 |
"n": 207,
|
| 9458 |
+
"num_samples": 1000
|
| 9459 |
},
|
| 9460 |
"n": 207
|
| 9461 |
}
|
workspace/runs/ctt_val_rollout_comparison/measured_metrics/report.md
CHANGED
|
@@ -5,23 +5,23 @@ K: `8`
|
|
| 5 |
| Metric | N | Micro mean | 95% CI | Task macro | Seed macro |
|
| 6 |
| --- | ---: | ---: | ---: | ---: | ---: |
|
| 7 |
| base_success | 207 | 0.2850 | [0.2271, 0.3478] | 0.2373 | 0.3227 |
|
| 8 |
-
| base_utility | 207 | 0.7358 | [0.
|
| 9 |
-
| branch_car_at_8 | 207 | 0.2936 | [0.
|
| 10 |
-
| hidden_chart_oracle_success_at_8 | 207 | 0.6667 | [0.
|
| 11 |
-
| hidden_chart_oracle_utility_at_8 | 207 | 1.4000 | [1.
|
| 12 |
-
| outcome_ptr_at_8 | 207 | 0.4589 | [0.3913, 0.
|
| 13 |
-
| pairwise_causal_calibration_ece | 207 | 0.1618 | [0.
|
| 14 |
-
| proposal_oracle_success_at_8 | 207 | 0.3768 | [0.3092, 0.
|
| 15 |
-
| proposal_oracle_success_gain_over_base_at_8 | 207 | 0.0918 | [0.
|
| 16 |
-
| proposal_oracle_utility_at_8 | 207 | 0.8811 | [0.
|
| 17 |
-
| proposal_oracle_utility_gain_over_base_at_8 | 207 | 0.1453 | [0.
|
| 18 |
| selected_success_at_8 | 207 | 0.2415 | [0.1884, 0.3043] | 0.2210 | 0.2842 |
|
| 19 |
-
| selected_success_gain_over_base_at_8 | 207 | -0.0435 | [-0.
|
| 20 |
-
| selected_utility_at_8 | 207 | 0.5875 | [0.
|
| 21 |
-
| selected_utility_gain_over_base_at_8 | 207 | -0.1483 | [-0.2654, -0.
|
| 22 |
-
| selector_regret_at_8 | 207 | 0.2936 | [0.
|
| 23 |
-
| success_selector_gap_at_8 | 207 | 0.1353 | [0.
|
| 24 |
-
| success_support_gap_at_8 | 207 | 0.2947 | [0.2319, 0.
|
| 25 |
| success_total_car_to_hidden_at_8 | 207 | 0.4300 | [0.3575, 0.4976] | 0.6456 | 0.5534 |
|
| 26 |
-
| support_gap_at_8 | 207 | 0.5283 | [0.4234, 0.
|
| 27 |
-
| total_car_to_hidden_at_8 | 207 | 0.8213 | [0.
|
|
|
|
| 5 |
| Metric | N | Micro mean | 95% CI | Task macro | Seed macro |
|
| 6 |
| --- | ---: | ---: | ---: | ---: | ---: |
|
| 7 |
| base_success | 207 | 0.2850 | [0.2271, 0.3478] | 0.2373 | 0.3227 |
|
| 8 |
+
| base_utility | 207 | 0.7358 | [0.6256, 0.8512] | 0.6892 | 0.8424 |
|
| 9 |
+
| branch_car_at_8 | 207 | 0.2936 | [0.2131, 0.3721] | 0.3799 | 0.4056 |
|
| 10 |
+
| hidden_chart_oracle_success_at_8 | 207 | 0.6667 | [0.5990, 0.7246] | 0.8647 | 0.8358 |
|
| 11 |
+
| hidden_chart_oracle_utility_at_8 | 207 | 1.4000 | [1.2800, 1.5063] | 1.7565 | 1.7062 |
|
| 12 |
+
| outcome_ptr_at_8 | 207 | 0.4589 | [0.3913, 0.5266] | 0.5411 | 0.4850 |
|
| 13 |
+
| pairwise_causal_calibration_ece | 207 | 0.1618 | [0.1463, 0.1771] | 0.1636 | 0.1579 |
|
| 14 |
+
| proposal_oracle_success_at_8 | 207 | 0.3768 | [0.3092, 0.4396] | 0.3800 | 0.4757 |
|
| 15 |
+
| proposal_oracle_success_gain_over_base_at_8 | 207 | 0.0918 | [0.0242, 0.1594] | 0.1427 | 0.1530 |
|
| 16 |
+
| proposal_oracle_utility_at_8 | 207 | 0.8811 | [0.7614, 1.0027] | 0.9713 | 1.1035 |
|
| 17 |
+
| proposal_oracle_utility_gain_over_base_at_8 | 207 | 0.1453 | [0.0309, 0.2609] | 0.2821 | 0.2611 |
|
| 18 |
| selected_success_at_8 | 207 | 0.2415 | [0.1884, 0.3043] | 0.2210 | 0.2842 |
|
| 19 |
+
| selected_success_gain_over_base_at_8 | 207 | -0.0435 | [-0.1063, 0.0242] | -0.0163 | -0.0385 |
|
| 20 |
+
| selected_utility_at_8 | 207 | 0.5875 | [0.4846, 0.7017] | 0.5914 | 0.6978 |
|
| 21 |
+
| selected_utility_gain_over_base_at_8 | 207 | -0.1483 | [-0.2654, -0.0240] | -0.0978 | -0.1446 |
|
| 22 |
+
| selector_regret_at_8 | 207 | 0.2936 | [0.2131, 0.3721] | 0.3799 | 0.4056 |
|
| 23 |
+
| success_selector_gap_at_8 | 207 | 0.1353 | [0.0918, 0.1836] | 0.1590 | 0.1915 |
|
| 24 |
+
| success_support_gap_at_8 | 207 | 0.2947 | [0.2319, 0.3575] | 0.4867 | 0.3619 |
|
| 25 |
| success_total_car_to_hidden_at_8 | 207 | 0.4300 | [0.3575, 0.4976] | 0.6456 | 0.5534 |
|
| 26 |
+
| support_gap_at_8 | 207 | 0.5283 | [0.4234, 0.6341] | 0.7890 | 0.6062 |
|
| 27 |
+
| total_car_to_hidden_at_8 | 207 | 0.8213 | [0.6984, 0.9381] | 1.1686 | 1.0116 |
|
workspace/runs/ctt_val_rollout_comparison/measured_metrics/table.tex
CHANGED
|
@@ -4,25 +4,25 @@
|
|
| 4 |
Metric & N & Micro mean & CI low & CI high \\
|
| 5 |
\midrule
|
| 6 |
base\_success & 207 & 0.2850 & 0.2271 & 0.3478 \\
|
| 7 |
-
base\_utility & 207 & 0.7358 & 0.
|
| 8 |
-
branch\_car\_at\_8 & 207 & 0.2936 & 0.
|
| 9 |
-
hidden\_chart\_oracle\_success\_at\_8 & 207 & 0.6667 & 0.
|
| 10 |
-
hidden\_chart\_oracle\_utility\_at\_8 & 207 & 1.4000 & 1.
|
| 11 |
-
outcome\_ptr\_at\_8 & 207 & 0.4589 & 0.3913 & 0.
|
| 12 |
-
pairwise\_causal\_calibration\_ece & 207 & 0.1618 & 0.
|
| 13 |
-
proposal\_oracle\_success\_at\_8 & 207 & 0.3768 & 0.3092 & 0.
|
| 14 |
-
proposal\_oracle\_success\_gain\_over\_base\_at\_8 & 207 & 0.0918 & 0.
|
| 15 |
-
proposal\_oracle\_utility\_at\_8 & 207 & 0.8811 & 0.
|
| 16 |
-
proposal\_oracle\_utility\_gain\_over\_base\_at\_8 & 207 & 0.1453 & 0.
|
| 17 |
selected\_success\_at\_8 & 207 & 0.2415 & 0.1884 & 0.3043 \\
|
| 18 |
-
selected\_success\_gain\_over\_base\_at\_8 & 207 & -0.0435 & -0.
|
| 19 |
-
selected\_utility\_at\_8 & 207 & 0.5875 & 0.
|
| 20 |
-
selected\_utility\_gain\_over\_base\_at\_8 & 207 & -0.1483 & -0.2654 & -0.
|
| 21 |
-
selector\_regret\_at\_8 & 207 & 0.2936 & 0.
|
| 22 |
-
success\_selector\_gap\_at\_8 & 207 & 0.1353 & 0.
|
| 23 |
-
success\_support\_gap\_at\_8 & 207 & 0.2947 & 0.2319 & 0.
|
| 24 |
success\_total\_car\_to\_hidden\_at\_8 & 207 & 0.4300 & 0.3575 & 0.4976 \\
|
| 25 |
-
support\_gap\_at\_8 & 207 & 0.5283 & 0.4234 & 0.
|
| 26 |
-
total\_car\_to\_hidden\_at\_8 & 207 & 0.8213 & 0.
|
| 27 |
\bottomrule
|
| 28 |
\end{tabular}
|
|
|
|
| 4 |
Metric & N & Micro mean & CI low & CI high \\
|
| 5 |
\midrule
|
| 6 |
base\_success & 207 & 0.2850 & 0.2271 & 0.3478 \\
|
| 7 |
+
base\_utility & 207 & 0.7358 & 0.6256 & 0.8512 \\
|
| 8 |
+
branch\_car\_at\_8 & 207 & 0.2936 & 0.2131 & 0.3721 \\
|
| 9 |
+
hidden\_chart\_oracle\_success\_at\_8 & 207 & 0.6667 & 0.5990 & 0.7246 \\
|
| 10 |
+
hidden\_chart\_oracle\_utility\_at\_8 & 207 & 1.4000 & 1.2800 & 1.5063 \\
|
| 11 |
+
outcome\_ptr\_at\_8 & 207 & 0.4589 & 0.3913 & 0.5266 \\
|
| 12 |
+
pairwise\_causal\_calibration\_ece & 207 & 0.1618 & 0.1463 & 0.1771 \\
|
| 13 |
+
proposal\_oracle\_success\_at\_8 & 207 & 0.3768 & 0.3092 & 0.4396 \\
|
| 14 |
+
proposal\_oracle\_success\_gain\_over\_base\_at\_8 & 207 & 0.0918 & 0.0242 & 0.1594 \\
|
| 15 |
+
proposal\_oracle\_utility\_at\_8 & 207 & 0.8811 & 0.7614 & 1.0027 \\
|
| 16 |
+
proposal\_oracle\_utility\_gain\_over\_base\_at\_8 & 207 & 0.1453 & 0.0309 & 0.2609 \\
|
| 17 |
selected\_success\_at\_8 & 207 & 0.2415 & 0.1884 & 0.3043 \\
|
| 18 |
+
selected\_success\_gain\_over\_base\_at\_8 & 207 & -0.0435 & -0.1063 & 0.0242 \\
|
| 19 |
+
selected\_utility\_at\_8 & 207 & 0.5875 & 0.4846 & 0.7017 \\
|
| 20 |
+
selected\_utility\_gain\_over\_base\_at\_8 & 207 & -0.1483 & -0.2654 & -0.0240 \\
|
| 21 |
+
selector\_regret\_at\_8 & 207 & 0.2936 & 0.2131 & 0.3721 \\
|
| 22 |
+
success\_selector\_gap\_at\_8 & 207 & 0.1353 & 0.0918 & 0.1836 \\
|
| 23 |
+
success\_support\_gap\_at\_8 & 207 & 0.2947 & 0.2319 & 0.3575 \\
|
| 24 |
success\_total\_car\_to\_hidden\_at\_8 & 207 & 0.4300 & 0.3575 & 0.4976 \\
|
| 25 |
+
support\_gap\_at\_8 & 207 & 0.5283 & 0.4234 & 0.6341 \\
|
| 26 |
+
total\_car\_to\_hidden\_at\_8 & 207 & 0.8213 & 0.6984 & 0.9381 \\
|
| 27 |
\bottomrule
|
| 28 |
\end{tabular}
|
workspace/runs/ctt_val_rollout_comparison/metrics.json
CHANGED
|
@@ -9,16 +9,25 @@
|
|
| 9 |
],
|
| 10 |
"source_content_hash": "90b431047a1f35cec095cbafa7de5bc0126cb0d0f4db3ff31843dca369977957",
|
| 11 |
"success_summary": {
|
|
|
|
|
|
|
| 12 |
"base_success_rate": 0.28502415458937197,
|
|
|
|
| 13 |
"base_utility_mean": 0.7358034186915535,
|
|
|
|
|
|
|
| 14 |
"hidden_chart_oracle_success_rate": 0.6666666666666666,
|
| 15 |
"hidden_chart_oracle_utility_mean": 1.3999846732054932,
|
| 16 |
"max_restore_error": 1.4901161193847656e-08,
|
|
|
|
| 17 |
"proposal_oracle_success_gain_over_base": 0.09178743961352658,
|
| 18 |
"proposal_oracle_success_rate": 0.37681159420289856,
|
|
|
|
| 19 |
"proposal_oracle_utility_mean": 0.8810875975700648,
|
|
|
|
| 20 |
"selected_success_gain_over_base": -0.043478260869565216,
|
| 21 |
"selected_success_rate": 0.24154589371980675,
|
|
|
|
| 22 |
"selected_utility_mean": 0.5875041471669958,
|
| 23 |
"success_selector_gap": 0.13526570048309178,
|
| 24 |
"success_support_gap": 0.2946859903381642
|
|
@@ -28,11 +37,11 @@
|
|
| 28 |
"macro_by_seed": {
|
| 29 |
"ci": {
|
| 30 |
"confidence": 0.95,
|
| 31 |
-
"high": 0.
|
| 32 |
-
"low": 0.
|
| 33 |
"mean": 0.322695035460993,
|
| 34 |
"n": 141,
|
| 35 |
-
"num_samples":
|
| 36 |
},
|
| 37 |
"mean": 0.322695035460993,
|
| 38 |
"per_seed": {
|
|
@@ -186,7 +195,7 @@
|
|
| 186 |
"low": 0.10666666666666666,
|
| 187 |
"mean": 0.23728506787330317,
|
| 188 |
"n": 5,
|
| 189 |
-
"num_samples":
|
| 190 |
},
|
| 191 |
"mean": 0.23728506787330317,
|
| 192 |
"per_task": {
|
|
@@ -204,7 +213,7 @@
|
|
| 204 |
"low": 0.22705314009661837,
|
| 205 |
"mean": 0.28502415458937197,
|
| 206 |
"n": 207,
|
| 207 |
-
"num_samples":
|
| 208 |
},
|
| 209 |
"n": 207
|
| 210 |
},
|
|
@@ -212,11 +221,11 @@
|
|
| 212 |
"macro_by_seed": {
|
| 213 |
"ci": {
|
| 214 |
"confidence": 0.95,
|
| 215 |
-
"high": 0.
|
| 216 |
-
"low": 0.
|
| 217 |
"mean": 0.84238802050422,
|
| 218 |
"n": 141,
|
| 219 |
-
"num_samples":
|
| 220 |
},
|
| 221 |
"mean": 0.84238802050422,
|
| 222 |
"per_seed": {
|
|
@@ -366,11 +375,11 @@
|
|
| 366 |
"macro_by_task": {
|
| 367 |
"ci": {
|
| 368 |
"confidence": 0.95,
|
| 369 |
-
"high": 0.
|
| 370 |
-
"low": 0.
|
| 371 |
"mean": 0.6891968629765195,
|
| 372 |
"n": 5,
|
| 373 |
-
"num_samples":
|
| 374 |
},
|
| 375 |
"mean": 0.6891968629765195,
|
| 376 |
"per_task": {
|
|
@@ -385,10 +394,10 @@
|
|
| 385 |
"micro": {
|
| 386 |
"confidence": 0.95,
|
| 387 |
"high": 0.851249745030154,
|
| 388 |
-
"low": 0.
|
| 389 |
"mean": 0.7358034186915535,
|
| 390 |
"n": 207,
|
| 391 |
-
"num_samples":
|
| 392 |
},
|
| 393 |
"n": 207
|
| 394 |
},
|
|
@@ -396,11 +405,11 @@
|
|
| 396 |
"macro_by_seed": {
|
| 397 |
"ci": {
|
| 398 |
"confidence": 0.95,
|
| 399 |
-
"high": 0.
|
| 400 |
-
"low": 0.
|
| 401 |
"mean": 0.40561739670197494,
|
| 402 |
"n": 141,
|
| 403 |
-
"num_samples":
|
| 404 |
},
|
| 405 |
"mean": 0.40561739670197494,
|
| 406 |
"per_seed": {
|
|
@@ -550,11 +559,11 @@
|
|
| 550 |
"macro_by_task": {
|
| 551 |
"ci": {
|
| 552 |
"confidence": 0.95,
|
| 553 |
-
"high": 0.
|
| 554 |
-
"low": 0.
|
| 555 |
"mean": 0.3798908010369334,
|
| 556 |
"n": 5,
|
| 557 |
-
"num_samples":
|
| 558 |
},
|
| 559 |
"mean": 0.3798908010369334,
|
| 560 |
"per_task": {
|
|
@@ -568,11 +577,11 @@
|
|
| 568 |
"metric": "branch_car_at_8",
|
| 569 |
"micro": {
|
| 570 |
"confidence": 0.95,
|
| 571 |
-
"high": 0.
|
| 572 |
-
"low": 0.
|
| 573 |
"mean": 0.293583450403069,
|
| 574 |
"n": 207,
|
| 575 |
-
"num_samples":
|
| 576 |
},
|
| 577 |
"n": 207
|
| 578 |
},
|
|
@@ -580,11 +589,11 @@
|
|
| 580 |
"macro_by_seed": {
|
| 581 |
"ci": {
|
| 582 |
"confidence": 0.95,
|
| 583 |
-
"high": 0.
|
| 584 |
-
"low": 0.
|
| 585 |
"mean": 0.8358156028368796,
|
| 586 |
"n": 141,
|
| 587 |
-
"num_samples":
|
| 588 |
},
|
| 589 |
"mean": 0.8358156028368796,
|
| 590 |
"per_seed": {
|
|
@@ -738,7 +747,7 @@
|
|
| 738 |
"low": 0.5941176470588235,
|
| 739 |
"mean": 0.8647058823529413,
|
| 740 |
"n": 5,
|
| 741 |
-
"num_samples":
|
| 742 |
},
|
| 743 |
"mean": 0.8647058823529413,
|
| 744 |
"per_task": {
|
|
@@ -753,10 +762,10 @@
|
|
| 753 |
"micro": {
|
| 754 |
"confidence": 0.95,
|
| 755 |
"high": 0.7246376811594203,
|
| 756 |
-
"low": 0.
|
| 757 |
"mean": 0.6666666666666666,
|
| 758 |
"n": 207,
|
| 759 |
-
"num_samples":
|
| 760 |
},
|
| 761 |
"n": 207
|
| 762 |
},
|
|
@@ -764,11 +773,11 @@
|
|
| 764 |
"macro_by_seed": {
|
| 765 |
"ci": {
|
| 766 |
"confidence": 0.95,
|
| 767 |
-
"high": 1.
|
| 768 |
-
"low": 1.
|
| 769 |
"mean": 1.706202811163896,
|
| 770 |
"n": 141,
|
| 771 |
-
"num_samples":
|
| 772 |
},
|
| 773 |
"mean": 1.706202811163896,
|
| 774 |
"per_seed": {
|
|
@@ -922,7 +931,7 @@
|
|
| 922 |
"low": 1.2693931020796299,
|
| 923 |
"mean": 1.7564643673598765,
|
| 924 |
"n": 5,
|
| 925 |
-
"num_samples":
|
| 926 |
},
|
| 927 |
"mean": 1.7564643673598765,
|
| 928 |
"per_task": {
|
|
@@ -936,11 +945,11 @@
|
|
| 936 |
"metric": "hidden_chart_oracle_utility_at_8",
|
| 937 |
"micro": {
|
| 938 |
"confidence": 0.95,
|
| 939 |
-
"high": 1.
|
| 940 |
-
"low": 1.
|
| 941 |
"mean": 1.3999846732054932,
|
| 942 |
"n": 207,
|
| 943 |
-
"num_samples":
|
| 944 |
},
|
| 945 |
"n": 207
|
| 946 |
},
|
|
@@ -948,11 +957,11 @@
|
|
| 948 |
"macro_by_seed": {
|
| 949 |
"ci": {
|
| 950 |
"confidence": 0.95,
|
| 951 |
-
"high": 0.
|
| 952 |
-
"low": 0.
|
| 953 |
"mean": 0.4849881796690307,
|
| 954 |
"n": 141,
|
| 955 |
-
"num_samples":
|
| 956 |
},
|
| 957 |
"mean": 0.4849881796690307,
|
| 958 |
"per_seed": {
|
|
@@ -1103,10 +1112,10 @@
|
|
| 1103 |
"ci": {
|
| 1104 |
"confidence": 0.95,
|
| 1105 |
"high": 0.6328205128205128,
|
| 1106 |
-
"low": 0.
|
| 1107 |
"mean": 0.5411161387631975,
|
| 1108 |
"n": 5,
|
| 1109 |
-
"num_samples":
|
| 1110 |
},
|
| 1111 |
"mean": 0.5411161387631975,
|
| 1112 |
"per_task": {
|
|
@@ -1120,11 +1129,11 @@
|
|
| 1120 |
"metric": "outcome_ptr_at_8",
|
| 1121 |
"micro": {
|
| 1122 |
"confidence": 0.95,
|
| 1123 |
-
"high": 0.
|
| 1124 |
"low": 0.391304347826087,
|
| 1125 |
"mean": 0.45893719806763283,
|
| 1126 |
"n": 207,
|
| 1127 |
-
"num_samples":
|
| 1128 |
},
|
| 1129 |
"n": 207
|
| 1130 |
},
|
|
@@ -1132,11 +1141,11 @@
|
|
| 1132 |
"macro_by_seed": {
|
| 1133 |
"ci": {
|
| 1134 |
"confidence": 0.95,
|
| 1135 |
-
"high": 0.
|
| 1136 |
-
"low": 0.
|
| 1137 |
"mean": 0.15793930902529071,
|
| 1138 |
"n": 141,
|
| 1139 |
-
"num_samples":
|
| 1140 |
},
|
| 1141 |
"mean": 0.15793930902529071,
|
| 1142 |
"per_seed": {
|
|
@@ -1286,11 +1295,11 @@
|
|
| 1286 |
"macro_by_task": {
|
| 1287 |
"ci": {
|
| 1288 |
"confidence": 0.95,
|
| 1289 |
-
"high": 0.
|
| 1290 |
-
"low": 0.
|
| 1291 |
"mean": 0.16359339232485098,
|
| 1292 |
"n": 5,
|
| 1293 |
-
"num_samples":
|
| 1294 |
},
|
| 1295 |
"mean": 0.16359339232485098,
|
| 1296 |
"per_task": {
|
|
@@ -1304,11 +1313,11 @@
|
|
| 1304 |
"metric": "pairwise_causal_calibration_ece",
|
| 1305 |
"micro": {
|
| 1306 |
"confidence": 0.95,
|
| 1307 |
-
"high": 0.
|
| 1308 |
-
"low": 0.
|
| 1309 |
"mean": 0.16176667684993604,
|
| 1310 |
"n": 207,
|
| 1311 |
-
"num_samples":
|
| 1312 |
},
|
| 1313 |
"n": 207
|
| 1314 |
},
|
|
@@ -1316,11 +1325,11 @@
|
|
| 1316 |
"macro_by_seed": {
|
| 1317 |
"ci": {
|
| 1318 |
"confidence": 0.95,
|
| 1319 |
-
"high": 0.
|
| 1320 |
-
"low": 0.
|
| 1321 |
"mean": 0.47565011820330966,
|
| 1322 |
"n": 141,
|
| 1323 |
-
"num_samples":
|
| 1324 |
},
|
| 1325 |
"mean": 0.47565011820330966,
|
| 1326 |
"per_seed": {
|
|
@@ -1471,10 +1480,10 @@
|
|
| 1471 |
"ci": {
|
| 1472 |
"confidence": 0.95,
|
| 1473 |
"high": 0.5987933634992458,
|
| 1474 |
-
"low": 0.
|
| 1475 |
"mean": 0.3799899446958271,
|
| 1476 |
"n": 5,
|
| 1477 |
-
"num_samples":
|
| 1478 |
},
|
| 1479 |
"mean": 0.3799899446958271,
|
| 1480 |
"per_task": {
|
|
@@ -1488,11 +1497,11 @@
|
|
| 1488 |
"metric": "proposal_oracle_success_at_8",
|
| 1489 |
"micro": {
|
| 1490 |
"confidence": 0.95,
|
| 1491 |
-
"high": 0.
|
| 1492 |
"low": 0.30917874396135264,
|
| 1493 |
"mean": 0.37681159420289856,
|
| 1494 |
"n": 207,
|
| 1495 |
-
"num_samples":
|
| 1496 |
},
|
| 1497 |
"n": 207
|
| 1498 |
},
|
|
@@ -1500,11 +1509,11 @@
|
|
| 1500 |
"macro_by_seed": {
|
| 1501 |
"ci": {
|
| 1502 |
"confidence": 0.95,
|
| 1503 |
-
"high": 0.
|
| 1504 |
-
"low": 0.
|
| 1505 |
"mean": 0.15295508274231676,
|
| 1506 |
"n": 141,
|
| 1507 |
-
"num_samples":
|
| 1508 |
},
|
| 1509 |
"mean": 0.15295508274231676,
|
| 1510 |
"per_seed": {
|
|
@@ -1658,7 +1667,7 @@
|
|
| 1658 |
"low": -0.01581699346405229,
|
| 1659 |
"mean": 0.1427048768225239,
|
| 1660 |
"n": 5,
|
| 1661 |
-
"num_samples":
|
| 1662 |
},
|
| 1663 |
"mean": 0.1427048768225239,
|
| 1664 |
"per_task": {
|
|
@@ -1673,10 +1682,10 @@
|
|
| 1673 |
"micro": {
|
| 1674 |
"confidence": 0.95,
|
| 1675 |
"high": 0.15942028985507245,
|
| 1676 |
-
"low": 0.
|
| 1677 |
"mean": 0.09178743961352658,
|
| 1678 |
"n": 207,
|
| 1679 |
-
"num_samples":
|
| 1680 |
},
|
| 1681 |
"n": 207
|
| 1682 |
},
|
|
@@ -1684,11 +1693,11 @@
|
|
| 1684 |
"macro_by_seed": {
|
| 1685 |
"ci": {
|
| 1686 |
"confidence": 0.95,
|
| 1687 |
-
"high": 1.
|
| 1688 |
-
"low": 0.
|
| 1689 |
"mean": 1.1034544456804372,
|
| 1690 |
"n": 141,
|
| 1691 |
-
"num_samples":
|
| 1692 |
},
|
| 1693 |
"mean": 1.1034544456804372,
|
| 1694 |
"per_seed": {
|
|
@@ -1839,10 +1848,10 @@
|
|
| 1839 |
"ci": {
|
| 1840 |
"confidence": 0.95,
|
| 1841 |
"high": 1.3557187332811518,
|
| 1842 |
-
"low": 0.
|
| 1843 |
"mean": 0.971318265720506,
|
| 1844 |
"n": 5,
|
| 1845 |
-
"num_samples":
|
| 1846 |
},
|
| 1847 |
"mean": 0.971318265720506,
|
| 1848 |
"per_task": {
|
|
@@ -1856,11 +1865,11 @@
|
|
| 1856 |
"metric": "proposal_oracle_utility_at_8",
|
| 1857 |
"micro": {
|
| 1858 |
"confidence": 0.95,
|
| 1859 |
-
"high": 1.
|
| 1860 |
-
"low": 0.
|
| 1861 |
"mean": 0.8810875975700648,
|
| 1862 |
"n": 207,
|
| 1863 |
-
"num_samples":
|
| 1864 |
},
|
| 1865 |
"n": 207
|
| 1866 |
},
|
|
@@ -1869,10 +1878,10 @@
|
|
| 1869 |
"ci": {
|
| 1870 |
"confidence": 0.95,
|
| 1871 |
"high": 0.40175874303166603,
|
| 1872 |
-
"low": 0.
|
| 1873 |
"mean": 0.2610664251762173,
|
| 1874 |
"n": 141,
|
| 1875 |
-
"num_samples":
|
| 1876 |
},
|
| 1877 |
"mean": 0.2610664251762173,
|
| 1878 |
"per_seed": {
|
|
@@ -2022,11 +2031,11 @@
|
|
| 2022 |
"macro_by_task": {
|
| 2023 |
"ci": {
|
| 2024 |
"confidence": 0.95,
|
| 2025 |
-
"high": 0.
|
| 2026 |
-
"low": -0.
|
| 2027 |
"mean": 0.28212140274398645,
|
| 2028 |
"n": 5,
|
| 2029 |
-
"num_samples":
|
| 2030 |
},
|
| 2031 |
"mean": 0.28212140274398645,
|
| 2032 |
"per_task": {
|
|
@@ -2040,11 +2049,11 @@
|
|
| 2040 |
"metric": "proposal_oracle_utility_gain_over_base_at_8",
|
| 2041 |
"micro": {
|
| 2042 |
"confidence": 0.95,
|
| 2043 |
-
"high": 0.
|
| 2044 |
-
"low": 0.
|
| 2045 |
"mean": 0.14528417887851136,
|
| 2046 |
"n": 207,
|
| 2047 |
-
"num_samples":
|
| 2048 |
},
|
| 2049 |
"n": 207
|
| 2050 |
},
|
|
@@ -2052,11 +2061,11 @@
|
|
| 2052 |
"macro_by_seed": {
|
| 2053 |
"ci": {
|
| 2054 |
"confidence": 0.95,
|
| 2055 |
-
"high": 0.
|
| 2056 |
-
"low": 0.
|
| 2057 |
"mean": 0.284160756501182,
|
| 2058 |
"n": 141,
|
| 2059 |
-
"num_samples":
|
| 2060 |
},
|
| 2061 |
"mean": 0.284160756501182,
|
| 2062 |
"per_seed": {
|
|
@@ -2206,11 +2215,11 @@
|
|
| 2206 |
"macro_by_task": {
|
| 2207 |
"ci": {
|
| 2208 |
"confidence": 0.95,
|
| 2209 |
-
"high": 0.
|
| 2210 |
"low": 0.07777777777777777,
|
| 2211 |
"mean": 0.22102564102564098,
|
| 2212 |
"n": 5,
|
| 2213 |
-
"num_samples":
|
| 2214 |
},
|
| 2215 |
"mean": 0.22102564102564098,
|
| 2216 |
"per_task": {
|
|
@@ -2228,7 +2237,7 @@
|
|
| 2228 |
"low": 0.18840579710144928,
|
| 2229 |
"mean": 0.24154589371980675,
|
| 2230 |
"n": 207,
|
| 2231 |
-
"num_samples":
|
| 2232 |
},
|
| 2233 |
"n": 207
|
| 2234 |
},
|
|
@@ -2236,11 +2245,11 @@
|
|
| 2236 |
"macro_by_seed": {
|
| 2237 |
"ci": {
|
| 2238 |
"confidence": 0.95,
|
| 2239 |
-
"high": 0.
|
| 2240 |
-
"low": -0.
|
| 2241 |
"mean": -0.038534278959810875,
|
| 2242 |
"n": 141,
|
| 2243 |
-
"num_samples":
|
| 2244 |
},
|
| 2245 |
"mean": -0.038534278959810875,
|
| 2246 |
"per_seed": {
|
|
@@ -2390,11 +2399,11 @@
|
|
| 2390 |
"macro_by_task": {
|
| 2391 |
"ci": {
|
| 2392 |
"confidence": 0.95,
|
| 2393 |
-
"high": 0.
|
| 2394 |
"low": -0.09986928104575163,
|
| 2395 |
"mean": -0.01625942684766214,
|
| 2396 |
"n": 5,
|
| 2397 |
-
"num_samples":
|
| 2398 |
},
|
| 2399 |
"mean": -0.01625942684766214,
|
| 2400 |
"per_task": {
|
|
@@ -2409,10 +2418,10 @@
|
|
| 2409 |
"micro": {
|
| 2410 |
"confidence": 0.95,
|
| 2411 |
"high": 0.024154589371980676,
|
| 2412 |
-
"low": -0.
|
| 2413 |
"mean": -0.043478260869565216,
|
| 2414 |
"n": 207,
|
| 2415 |
-
"num_samples":
|
| 2416 |
},
|
| 2417 |
"n": 207
|
| 2418 |
},
|
|
@@ -2420,11 +2429,11 @@
|
|
| 2420 |
"macro_by_seed": {
|
| 2421 |
"ci": {
|
| 2422 |
"confidence": 0.95,
|
| 2423 |
-
"high": 0.
|
| 2424 |
-
"low": 0.
|
| 2425 |
"mean": 0.6978370489784622,
|
| 2426 |
"n": 141,
|
| 2427 |
-
"num_samples":
|
| 2428 |
},
|
| 2429 |
"mean": 0.6978370489784622,
|
| 2430 |
"per_seed": {
|
|
@@ -2575,10 +2584,10 @@
|
|
| 2575 |
"ci": {
|
| 2576 |
"confidence": 0.95,
|
| 2577 |
"high": 0.886751612873159,
|
| 2578 |
-
"low": 0.
|
| 2579 |
"mean": 0.5914274646835725,
|
| 2580 |
"n": 5,
|
| 2581 |
-
"num_samples":
|
| 2582 |
},
|
| 2583 |
"mean": 0.5914274646835725,
|
| 2584 |
"per_task": {
|
|
@@ -2592,11 +2601,11 @@
|
|
| 2592 |
"metric": "selected_utility_at_8",
|
| 2593 |
"micro": {
|
| 2594 |
"confidence": 0.95,
|
| 2595 |
-
"high": 0.
|
| 2596 |
-
"low": 0.
|
| 2597 |
"mean": 0.5875041471669958,
|
| 2598 |
"n": 207,
|
| 2599 |
-
"num_samples":
|
| 2600 |
},
|
| 2601 |
"n": 207
|
| 2602 |
},
|
|
@@ -2604,11 +2613,11 @@
|
|
| 2604 |
"macro_by_seed": {
|
| 2605 |
"ci": {
|
| 2606 |
"confidence": 0.95,
|
| 2607 |
-
"high": 0.
|
| 2608 |
-
"low": -0.
|
| 2609 |
"mean": -0.14455097152575763,
|
| 2610 |
"n": 141,
|
| 2611 |
-
"num_samples":
|
| 2612 |
},
|
| 2613 |
"mean": -0.14455097152575763,
|
| 2614 |
"per_seed": {
|
|
@@ -2758,11 +2767,11 @@
|
|
| 2758 |
"macro_by_task": {
|
| 2759 |
"ci": {
|
| 2760 |
"confidence": 0.95,
|
| 2761 |
-
"high": 0.
|
| 2762 |
"low": -0.2553936764363997,
|
| 2763 |
"mean": -0.09776939829294695,
|
| 2764 |
"n": 5,
|
| 2765 |
-
"num_samples":
|
| 2766 |
},
|
| 2767 |
"mean": -0.09776939829294695,
|
| 2768 |
"per_task": {
|
|
@@ -2776,11 +2785,11 @@
|
|
| 2776 |
"metric": "selected_utility_gain_over_base_at_8",
|
| 2777 |
"micro": {
|
| 2778 |
"confidence": 0.95,
|
| 2779 |
-
"high": -0.
|
| 2780 |
"low": -0.26539749033037113,
|
| 2781 |
"mean": -0.14829927152455763,
|
| 2782 |
"n": 207,
|
| 2783 |
-
"num_samples":
|
| 2784 |
},
|
| 2785 |
"n": 207
|
| 2786 |
},
|
|
@@ -2788,11 +2797,11 @@
|
|
| 2788 |
"macro_by_seed": {
|
| 2789 |
"ci": {
|
| 2790 |
"confidence": 0.95,
|
| 2791 |
-
"high": 0.
|
| 2792 |
-
"low": 0.
|
| 2793 |
"mean": 0.40561739670197494,
|
| 2794 |
"n": 141,
|
| 2795 |
-
"num_samples":
|
| 2796 |
},
|
| 2797 |
"mean": 0.40561739670197494,
|
| 2798 |
"per_seed": {
|
|
@@ -2942,11 +2951,11 @@
|
|
| 2942 |
"macro_by_task": {
|
| 2943 |
"ci": {
|
| 2944 |
"confidence": 0.95,
|
| 2945 |
-
"high": 0.
|
| 2946 |
-
"low": 0.
|
| 2947 |
"mean": 0.3798908010369334,
|
| 2948 |
"n": 5,
|
| 2949 |
-
"num_samples":
|
| 2950 |
},
|
| 2951 |
"mean": 0.3798908010369334,
|
| 2952 |
"per_task": {
|
|
@@ -2960,11 +2969,11 @@
|
|
| 2960 |
"metric": "selector_regret_at_8",
|
| 2961 |
"micro": {
|
| 2962 |
"confidence": 0.95,
|
| 2963 |
-
"high": 0.
|
| 2964 |
-
"low": 0.
|
| 2965 |
"mean": 0.293583450403069,
|
| 2966 |
"n": 207,
|
| 2967 |
-
"num_samples":
|
| 2968 |
},
|
| 2969 |
"n": 207
|
| 2970 |
},
|
|
@@ -2972,11 +2981,11 @@
|
|
| 2972 |
"macro_by_seed": {
|
| 2973 |
"ci": {
|
| 2974 |
"confidence": 0.95,
|
| 2975 |
-
"high": 0.
|
| 2976 |
"low": 0.1276595744680851,
|
| 2977 |
"mean": 0.19148936170212766,
|
| 2978 |
"n": 141,
|
| 2979 |
-
"num_samples":
|
| 2980 |
},
|
| 2981 |
"mean": 0.19148936170212766,
|
| 2982 |
"per_seed": {
|
|
@@ -3126,11 +3135,11 @@
|
|
| 3126 |
"macro_by_task": {
|
| 3127 |
"ci": {
|
| 3128 |
"confidence": 0.95,
|
| 3129 |
-
"high": 0.
|
| 3130 |
-
"low": 0.
|
| 3131 |
"mean": 0.15896430367018605,
|
| 3132 |
"n": 5,
|
| 3133 |
-
"num_samples":
|
| 3134 |
},
|
| 3135 |
"mean": 0.15896430367018605,
|
| 3136 |
"per_task": {
|
|
@@ -3144,11 +3153,11 @@
|
|
| 3144 |
"metric": "success_selector_gap_at_8",
|
| 3145 |
"micro": {
|
| 3146 |
"confidence": 0.95,
|
| 3147 |
-
"high": 0.
|
| 3148 |
-
"low": 0.
|
| 3149 |
"mean": 0.13526570048309178,
|
| 3150 |
"n": 207,
|
| 3151 |
-
"num_samples":
|
| 3152 |
},
|
| 3153 |
"n": 207
|
| 3154 |
},
|
|
@@ -3156,11 +3165,11 @@
|
|
| 3156 |
"macro_by_seed": {
|
| 3157 |
"ci": {
|
| 3158 |
"confidence": 0.95,
|
| 3159 |
-
"high": 0.
|
| 3160 |
-
"low": 0.
|
| 3161 |
"mean": 0.3619385342789597,
|
| 3162 |
"n": 141,
|
| 3163 |
-
"num_samples":
|
| 3164 |
},
|
| 3165 |
"mean": 0.3619385342789597,
|
| 3166 |
"per_seed": {
|
|
@@ -3314,7 +3323,7 @@
|
|
| 3314 |
"low": 0.24550025138260434,
|
| 3315 |
"mean": 0.48667672197083967,
|
| 3316 |
"n": 5,
|
| 3317 |
-
"num_samples":
|
| 3318 |
},
|
| 3319 |
"mean": 0.48667672197083967,
|
| 3320 |
"per_task": {
|
|
@@ -3328,11 +3337,11 @@
|
|
| 3328 |
"metric": "success_support_gap_at_8",
|
| 3329 |
"micro": {
|
| 3330 |
"confidence": 0.95,
|
| 3331 |
-
"high": 0.
|
| 3332 |
"low": 0.2318840579710145,
|
| 3333 |
"mean": 0.2946859903381642,
|
| 3334 |
"n": 207,
|
| 3335 |
-
"num_samples":
|
| 3336 |
},
|
| 3337 |
"n": 207
|
| 3338 |
},
|
|
@@ -3340,11 +3349,11 @@
|
|
| 3340 |
"macro_by_seed": {
|
| 3341 |
"ci": {
|
| 3342 |
"confidence": 0.95,
|
| 3343 |
-
"high": 0.
|
| 3344 |
-
"low": 0.
|
| 3345 |
"mean": 0.5534278959810874,
|
| 3346 |
"n": 141,
|
| 3347 |
-
"num_samples":
|
| 3348 |
},
|
| 3349 |
"mean": 0.5534278959810874,
|
| 3350 |
"per_seed": {
|
|
@@ -3494,11 +3503,11 @@
|
|
| 3494 |
"macro_by_task": {
|
| 3495 |
"ci": {
|
| 3496 |
"confidence": 0.95,
|
| 3497 |
-
"high": 0.
|
| 3498 |
-
"low": 0.
|
| 3499 |
"mean": 0.6456410256410257,
|
| 3500 |
"n": 5,
|
| 3501 |
-
"num_samples":
|
| 3502 |
},
|
| 3503 |
"mean": 0.6456410256410257,
|
| 3504 |
"per_task": {
|
|
@@ -3516,7 +3525,7 @@
|
|
| 3516 |
"low": 0.357487922705314,
|
| 3517 |
"mean": 0.42995169082125606,
|
| 3518 |
"n": 207,
|
| 3519 |
-
"num_samples":
|
| 3520 |
},
|
| 3521 |
"n": 207
|
| 3522 |
},
|
|
@@ -3524,11 +3533,11 @@
|
|
| 3524 |
"macro_by_seed": {
|
| 3525 |
"ci": {
|
| 3526 |
"confidence": 0.95,
|
| 3527 |
-
"high": 0.
|
| 3528 |
-
"low": 0.
|
| 3529 |
"mean": 0.6062044484887313,
|
| 3530 |
"n": 141,
|
| 3531 |
-
"num_samples":
|
| 3532 |
},
|
| 3533 |
"mean": 0.6062044484887313,
|
| 3534 |
"per_seed": {
|
|
@@ -3682,7 +3691,7 @@
|
|
| 3682 |
"low": 0.4585766474838634,
|
| 3683 |
"mean": 0.788961660718036,
|
| 3684 |
"n": 5,
|
| 3685 |
-
"num_samples":
|
| 3686 |
},
|
| 3687 |
"mean": 0.788961660718036,
|
| 3688 |
"per_task": {
|
|
@@ -3696,11 +3705,11 @@
|
|
| 3696 |
"metric": "support_gap_at_8",
|
| 3697 |
"micro": {
|
| 3698 |
"confidence": 0.95,
|
| 3699 |
-
"high": 0.
|
| 3700 |
"low": 0.4234259423213799,
|
| 3701 |
"mean": 0.5282977284379371,
|
| 3702 |
"n": 207,
|
| 3703 |
-
"num_samples":
|
| 3704 |
},
|
| 3705 |
"n": 207
|
| 3706 |
},
|
|
@@ -3708,11 +3717,11 @@
|
|
| 3708 |
"macro_by_seed": {
|
| 3709 |
"ci": {
|
| 3710 |
"confidence": 0.95,
|
| 3711 |
-
"high": 1.
|
| 3712 |
-
"low": 0.
|
| 3713 |
"mean": 1.0115917835033712,
|
| 3714 |
"n": 141,
|
| 3715 |
-
"num_samples":
|
| 3716 |
},
|
| 3717 |
"mean": 1.0115917835033712,
|
| 3718 |
"per_seed": {
|
|
@@ -3866,7 +3875,7 @@
|
|
| 3866 |
"low": 0.722751199278172,
|
| 3867 |
"mean": 1.1686045027220224,
|
| 3868 |
"n": 5,
|
| 3869 |
-
"num_samples":
|
| 3870 |
},
|
| 3871 |
"mean": 1.1686045027220224,
|
| 3872 |
"per_task": {
|
|
@@ -3880,11 +3889,11 @@
|
|
| 3880 |
"metric": "total_car_to_hidden_at_8",
|
| 3881 |
"micro": {
|
| 3882 |
"confidence": 0.95,
|
| 3883 |
-
"high": 0.
|
| 3884 |
-
"low": 0.
|
| 3885 |
"mean": 0.8212702652815715,
|
| 3886 |
"n": 207,
|
| 3887 |
-
"num_samples":
|
| 3888 |
},
|
| 3889 |
"n": 207
|
| 3890 |
}
|
|
|
|
| 9 |
],
|
| 10 |
"source_content_hash": "90b431047a1f35cec095cbafa7de5bc0126cb0d0f4db3ff31843dca369977957",
|
| 11 |
"success_summary": {
|
| 12 |
+
"any_generated_unsafe_known": null,
|
| 13 |
+
"base_safety_label_known_rate": null,
|
| 14 |
"base_success_rate": 0.28502415458937197,
|
| 15 |
+
"base_unsafe_rate_known": null,
|
| 16 |
"base_utility_mean": 0.7358034186915535,
|
| 17 |
+
"generated_safety_label_coverage": null,
|
| 18 |
+
"generated_unsafe_rate_known": null,
|
| 19 |
"hidden_chart_oracle_success_rate": 0.6666666666666666,
|
| 20 |
"hidden_chart_oracle_utility_mean": 1.3999846732054932,
|
| 21 |
"max_restore_error": 1.4901161193847656e-08,
|
| 22 |
+
"proposal_oracle_safety_label_known_rate": null,
|
| 23 |
"proposal_oracle_success_gain_over_base": 0.09178743961352658,
|
| 24 |
"proposal_oracle_success_rate": 0.37681159420289856,
|
| 25 |
+
"proposal_oracle_unsafe_rate_known": null,
|
| 26 |
"proposal_oracle_utility_mean": 0.8810875975700648,
|
| 27 |
+
"selected_safety_label_known_rate": null,
|
| 28 |
"selected_success_gain_over_base": -0.043478260869565216,
|
| 29 |
"selected_success_rate": 0.24154589371980675,
|
| 30 |
+
"selected_unsafe_rate_known": null,
|
| 31 |
"selected_utility_mean": 0.5875041471669958,
|
| 32 |
"success_selector_gap": 0.13526570048309178,
|
| 33 |
"success_support_gap": 0.2946859903381642
|
|
|
|
| 37 |
"macro_by_seed": {
|
| 38 |
"ci": {
|
| 39 |
"confidence": 0.95,
|
| 40 |
+
"high": 0.3918439716312056,
|
| 41 |
+
"low": 0.25,
|
| 42 |
"mean": 0.322695035460993,
|
| 43 |
"n": 141,
|
| 44 |
+
"num_samples": 1000
|
| 45 |
},
|
| 46 |
"mean": 0.322695035460993,
|
| 47 |
"per_seed": {
|
|
|
|
| 195 |
"low": 0.10666666666666666,
|
| 196 |
"mean": 0.23728506787330317,
|
| 197 |
"n": 5,
|
| 198 |
+
"num_samples": 1000
|
| 199 |
},
|
| 200 |
"mean": 0.23728506787330317,
|
| 201 |
"per_task": {
|
|
|
|
| 213 |
"low": 0.22705314009661837,
|
| 214 |
"mean": 0.28502415458937197,
|
| 215 |
"n": 207,
|
| 216 |
+
"num_samples": 1000
|
| 217 |
},
|
| 218 |
"n": 207
|
| 219 |
},
|
|
|
|
| 221 |
"macro_by_seed": {
|
| 222 |
"ci": {
|
| 223 |
"confidence": 0.95,
|
| 224 |
+
"high": 0.9671404774753207,
|
| 225 |
+
"low": 0.7101003489727332,
|
| 226 |
"mean": 0.84238802050422,
|
| 227 |
"n": 141,
|
| 228 |
+
"num_samples": 1000
|
| 229 |
},
|
| 230 |
"mean": 0.84238802050422,
|
| 231 |
"per_seed": {
|
|
|
|
| 375 |
"macro_by_task": {
|
| 376 |
"ci": {
|
| 377 |
"confidence": 0.95,
|
| 378 |
+
"high": 0.924619877206718,
|
| 379 |
+
"low": 0.40042830170729216,
|
| 380 |
"mean": 0.6891968629765195,
|
| 381 |
"n": 5,
|
| 382 |
+
"num_samples": 1000
|
| 383 |
},
|
| 384 |
"mean": 0.6891968629765195,
|
| 385 |
"per_task": {
|
|
|
|
| 394 |
"micro": {
|
| 395 |
"confidence": 0.95,
|
| 396 |
"high": 0.851249745030154,
|
| 397 |
+
"low": 0.625559344417573,
|
| 398 |
"mean": 0.7358034186915535,
|
| 399 |
"n": 207,
|
| 400 |
+
"num_samples": 1000
|
| 401 |
},
|
| 402 |
"n": 207
|
| 403 |
},
|
|
|
|
| 405 |
"macro_by_seed": {
|
| 406 |
"ci": {
|
| 407 |
"confidence": 0.95,
|
| 408 |
+
"high": 0.5194647902708409,
|
| 409 |
+
"low": 0.2999603762294949,
|
| 410 |
"mean": 0.40561739670197494,
|
| 411 |
"n": 141,
|
| 412 |
+
"num_samples": 1000
|
| 413 |
},
|
| 414 |
"mean": 0.40561739670197494,
|
| 415 |
"per_seed": {
|
|
|
|
| 559 |
"macro_by_task": {
|
| 560 |
"ci": {
|
| 561 |
"confidence": 0.95,
|
| 562 |
+
"high": 0.5319262530406317,
|
| 563 |
+
"low": 0.19367229081819662,
|
| 564 |
"mean": 0.3798908010369334,
|
| 565 |
"n": 5,
|
| 566 |
+
"num_samples": 1000
|
| 567 |
},
|
| 568 |
"mean": 0.3798908010369334,
|
| 569 |
"per_task": {
|
|
|
|
| 577 |
"metric": "branch_car_at_8",
|
| 578 |
"micro": {
|
| 579 |
"confidence": 0.95,
|
| 580 |
+
"high": 0.3721372773162211,
|
| 581 |
+
"low": 0.21313571063115527,
|
| 582 |
"mean": 0.293583450403069,
|
| 583 |
"n": 207,
|
| 584 |
+
"num_samples": 1000
|
| 585 |
},
|
| 586 |
"n": 207
|
| 587 |
},
|
|
|
|
| 589 |
"macro_by_seed": {
|
| 590 |
"ci": {
|
| 591 |
"confidence": 0.95,
|
| 592 |
+
"high": 0.8887706855791963,
|
| 593 |
+
"low": 0.7754137115839246,
|
| 594 |
"mean": 0.8358156028368796,
|
| 595 |
"n": 141,
|
| 596 |
+
"num_samples": 1000
|
| 597 |
},
|
| 598 |
"mean": 0.8358156028368796,
|
| 599 |
"per_seed": {
|
|
|
|
| 747 |
"low": 0.5941176470588235,
|
| 748 |
"mean": 0.8647058823529413,
|
| 749 |
"n": 5,
|
| 750 |
+
"num_samples": 1000
|
| 751 |
},
|
| 752 |
"mean": 0.8647058823529413,
|
| 753 |
"per_task": {
|
|
|
|
| 762 |
"micro": {
|
| 763 |
"confidence": 0.95,
|
| 764 |
"high": 0.7246376811594203,
|
| 765 |
+
"low": 0.5990338164251208,
|
| 766 |
"mean": 0.6666666666666666,
|
| 767 |
"n": 207,
|
| 768 |
+
"num_samples": 1000
|
| 769 |
},
|
| 770 |
"n": 207
|
| 771 |
},
|
|
|
|
| 773 |
"macro_by_seed": {
|
| 774 |
"ci": {
|
| 775 |
"confidence": 0.95,
|
| 776 |
+
"high": 1.8010394367889166,
|
| 777 |
+
"low": 1.5995275395488624,
|
| 778 |
"mean": 1.706202811163896,
|
| 779 |
"n": 141,
|
| 780 |
+
"num_samples": 1000
|
| 781 |
},
|
| 782 |
"mean": 1.706202811163896,
|
| 783 |
"per_seed": {
|
|
|
|
| 931 |
"low": 1.2693931020796299,
|
| 932 |
"mean": 1.7564643673598765,
|
| 933 |
"n": 5,
|
| 934 |
+
"num_samples": 1000
|
| 935 |
},
|
| 936 |
"mean": 1.7564643673598765,
|
| 937 |
"per_task": {
|
|
|
|
| 945 |
"metric": "hidden_chart_oracle_utility_at_8",
|
| 946 |
"micro": {
|
| 947 |
"confidence": 0.95,
|
| 948 |
+
"high": 1.5062634588583657,
|
| 949 |
+
"low": 1.2799855626507657,
|
| 950 |
"mean": 1.3999846732054932,
|
| 951 |
"n": 207,
|
| 952 |
+
"num_samples": 1000
|
| 953 |
},
|
| 954 |
"n": 207
|
| 955 |
},
|
|
|
|
| 957 |
"macro_by_seed": {
|
| 958 |
"ci": {
|
| 959 |
"confidence": 0.95,
|
| 960 |
+
"high": 0.5674940898345152,
|
| 961 |
+
"low": 0.4113475177304965,
|
| 962 |
"mean": 0.4849881796690307,
|
| 963 |
"n": 141,
|
| 964 |
+
"num_samples": 1000
|
| 965 |
},
|
| 966 |
"mean": 0.4849881796690307,
|
| 967 |
"per_seed": {
|
|
|
|
| 1112 |
"ci": {
|
| 1113 |
"confidence": 0.95,
|
| 1114 |
"high": 0.6328205128205128,
|
| 1115 |
+
"low": 0.4294117647058823,
|
| 1116 |
"mean": 0.5411161387631975,
|
| 1117 |
"n": 5,
|
| 1118 |
+
"num_samples": 1000
|
| 1119 |
},
|
| 1120 |
"mean": 0.5411161387631975,
|
| 1121 |
"per_task": {
|
|
|
|
| 1129 |
"metric": "outcome_ptr_at_8",
|
| 1130 |
"micro": {
|
| 1131 |
"confidence": 0.95,
|
| 1132 |
+
"high": 0.5265700483091788,
|
| 1133 |
"low": 0.391304347826087,
|
| 1134 |
"mean": 0.45893719806763283,
|
| 1135 |
"n": 207,
|
| 1136 |
+
"num_samples": 1000
|
| 1137 |
},
|
| 1138 |
"n": 207
|
| 1139 |
},
|
|
|
|
| 1141 |
"macro_by_seed": {
|
| 1142 |
"ci": {
|
| 1143 |
"confidence": 0.95,
|
| 1144 |
+
"high": 0.17446188421793687,
|
| 1145 |
+
"low": 0.14114248449490324,
|
| 1146 |
"mean": 0.15793930902529071,
|
| 1147 |
"n": 141,
|
| 1148 |
+
"num_samples": 1000
|
| 1149 |
},
|
| 1150 |
"mean": 0.15793930902529071,
|
| 1151 |
"per_seed": {
|
|
|
|
| 1295 |
"macro_by_task": {
|
| 1296 |
"ci": {
|
| 1297 |
"confidence": 0.95,
|
| 1298 |
+
"high": 0.17042242389065887,
|
| 1299 |
+
"low": 0.1574042611063619,
|
| 1300 |
"mean": 0.16359339232485098,
|
| 1301 |
"n": 5,
|
| 1302 |
+
"num_samples": 1000
|
| 1303 |
},
|
| 1304 |
"mean": 0.16359339232485098,
|
| 1305 |
"per_task": {
|
|
|
|
| 1313 |
"metric": "pairwise_causal_calibration_ece",
|
| 1314 |
"micro": {
|
| 1315 |
"confidence": 0.95,
|
| 1316 |
+
"high": 0.1771431506285629,
|
| 1317 |
+
"low": 0.14633605302901057,
|
| 1318 |
"mean": 0.16176667684993604,
|
| 1319 |
"n": 207,
|
| 1320 |
+
"num_samples": 1000
|
| 1321 |
},
|
| 1322 |
"n": 207
|
| 1323 |
},
|
|
|
|
| 1325 |
"macro_by_seed": {
|
| 1326 |
"ci": {
|
| 1327 |
"confidence": 0.95,
|
| 1328 |
+
"high": 0.5508274231678488,
|
| 1329 |
+
"low": 0.3905437352245863,
|
| 1330 |
"mean": 0.47565011820330966,
|
| 1331 |
"n": 141,
|
| 1332 |
+
"num_samples": 1000
|
| 1333 |
},
|
| 1334 |
"mean": 0.47565011820330966,
|
| 1335 |
"per_seed": {
|
|
|
|
| 1480 |
"ci": {
|
| 1481 |
"confidence": 0.95,
|
| 1482 |
"high": 0.5987933634992458,
|
| 1483 |
+
"low": 0.13725490196078433,
|
| 1484 |
"mean": 0.3799899446958271,
|
| 1485 |
"n": 5,
|
| 1486 |
+
"num_samples": 1000
|
| 1487 |
},
|
| 1488 |
"mean": 0.3799899446958271,
|
| 1489 |
"per_task": {
|
|
|
|
| 1497 |
"metric": "proposal_oracle_success_at_8",
|
| 1498 |
"micro": {
|
| 1499 |
"confidence": 0.95,
|
| 1500 |
+
"high": 0.4396135265700483,
|
| 1501 |
"low": 0.30917874396135264,
|
| 1502 |
"mean": 0.37681159420289856,
|
| 1503 |
"n": 207,
|
| 1504 |
+
"num_samples": 1000
|
| 1505 |
},
|
| 1506 |
"n": 207
|
| 1507 |
},
|
|
|
|
| 1509 |
"macro_by_seed": {
|
| 1510 |
"ci": {
|
| 1511 |
"confidence": 0.95,
|
| 1512 |
+
"high": 0.23179669030732858,
|
| 1513 |
+
"low": 0.06643026004728132,
|
| 1514 |
"mean": 0.15295508274231676,
|
| 1515 |
"n": 141,
|
| 1516 |
+
"num_samples": 1000
|
| 1517 |
},
|
| 1518 |
"mean": 0.15295508274231676,
|
| 1519 |
"per_seed": {
|
|
|
|
| 1667 |
"low": -0.01581699346405229,
|
| 1668 |
"mean": 0.1427048768225239,
|
| 1669 |
"n": 5,
|
| 1670 |
+
"num_samples": 1000
|
| 1671 |
},
|
| 1672 |
"mean": 0.1427048768225239,
|
| 1673 |
"per_task": {
|
|
|
|
| 1682 |
"micro": {
|
| 1683 |
"confidence": 0.95,
|
| 1684 |
"high": 0.15942028985507245,
|
| 1685 |
+
"low": 0.024154589371980676,
|
| 1686 |
"mean": 0.09178743961352658,
|
| 1687 |
"n": 207,
|
| 1688 |
+
"num_samples": 1000
|
| 1689 |
},
|
| 1690 |
"n": 207
|
| 1691 |
},
|
|
|
|
| 1693 |
"macro_by_seed": {
|
| 1694 |
"ci": {
|
| 1695 |
"confidence": 0.95,
|
| 1696 |
+
"high": 1.2371345245170386,
|
| 1697 |
+
"low": 0.9570726445954775,
|
| 1698 |
"mean": 1.1034544456804372,
|
| 1699 |
"n": 141,
|
| 1700 |
+
"num_samples": 1000
|
| 1701 |
},
|
| 1702 |
"mean": 1.1034544456804372,
|
| 1703 |
"per_seed": {
|
|
|
|
| 1848 |
"ci": {
|
| 1849 |
"confidence": 0.95,
|
| 1850 |
"high": 1.3557187332811518,
|
| 1851 |
+
"low": 0.5600978099637561,
|
| 1852 |
"mean": 0.971318265720506,
|
| 1853 |
"n": 5,
|
| 1854 |
+
"num_samples": 1000
|
| 1855 |
},
|
| 1856 |
"mean": 0.971318265720506,
|
| 1857 |
"per_task": {
|
|
|
|
| 1865 |
"metric": "proposal_oracle_utility_at_8",
|
| 1866 |
"micro": {
|
| 1867 |
"confidence": 0.95,
|
| 1868 |
+
"high": 1.0026757181962336,
|
| 1869 |
+
"low": 0.7613600684022801,
|
| 1870 |
"mean": 0.8810875975700648,
|
| 1871 |
"n": 207,
|
| 1872 |
+
"num_samples": 1000
|
| 1873 |
},
|
| 1874 |
"n": 207
|
| 1875 |
},
|
|
|
|
| 1878 |
"ci": {
|
| 1879 |
"confidence": 0.95,
|
| 1880 |
"high": 0.40175874303166603,
|
| 1881 |
+
"low": 0.11287717537031931,
|
| 1882 |
"mean": 0.2610664251762173,
|
| 1883 |
"n": 141,
|
| 1884 |
+
"num_samples": 1000
|
| 1885 |
},
|
| 1886 |
"mean": 0.2610664251762173,
|
| 1887 |
"per_seed": {
|
|
|
|
| 2031 |
"macro_by_task": {
|
| 2032 |
"ci": {
|
| 2033 |
"confidence": 0.95,
|
| 2034 |
+
"high": 0.5660580151865625,
|
| 2035 |
+
"low": -0.016164922901372936,
|
| 2036 |
"mean": 0.28212140274398645,
|
| 2037 |
"n": 5,
|
| 2038 |
+
"num_samples": 1000
|
| 2039 |
},
|
| 2040 |
"mean": 0.28212140274398645,
|
| 2041 |
"per_task": {
|
|
|
|
| 2049 |
"metric": "proposal_oracle_utility_gain_over_base_at_8",
|
| 2050 |
"micro": {
|
| 2051 |
"confidence": 0.95,
|
| 2052 |
+
"high": 0.2609384466434496,
|
| 2053 |
+
"low": 0.03087088158933428,
|
| 2054 |
"mean": 0.14528417887851136,
|
| 2055 |
"n": 207,
|
| 2056 |
+
"num_samples": 1000
|
| 2057 |
},
|
| 2058 |
"n": 207
|
| 2059 |
},
|
|
|
|
| 2061 |
"macro_by_seed": {
|
| 2062 |
"ci": {
|
| 2063 |
"confidence": 0.95,
|
| 2064 |
+
"high": 0.35579196217494086,
|
| 2065 |
+
"low": 0.2118203309692671,
|
| 2066 |
"mean": 0.284160756501182,
|
| 2067 |
"n": 141,
|
| 2068 |
+
"num_samples": 1000
|
| 2069 |
},
|
| 2070 |
"mean": 0.284160756501182,
|
| 2071 |
"per_seed": {
|
|
|
|
| 2215 |
"macro_by_task": {
|
| 2216 |
"ci": {
|
| 2217 |
"confidence": 0.95,
|
| 2218 |
+
"high": 0.391965811965812,
|
| 2219 |
"low": 0.07777777777777777,
|
| 2220 |
"mean": 0.22102564102564098,
|
| 2221 |
"n": 5,
|
| 2222 |
+
"num_samples": 1000
|
| 2223 |
},
|
| 2224 |
"mean": 0.22102564102564098,
|
| 2225 |
"per_task": {
|
|
|
|
| 2237 |
"low": 0.18840579710144928,
|
| 2238 |
"mean": 0.24154589371980675,
|
| 2239 |
"n": 207,
|
| 2240 |
+
"num_samples": 1000
|
| 2241 |
},
|
| 2242 |
"n": 207
|
| 2243 |
},
|
|
|
|
| 2245 |
"macro_by_seed": {
|
| 2246 |
"ci": {
|
| 2247 |
"confidence": 0.95,
|
| 2248 |
+
"high": 0.04893617021276596,
|
| 2249 |
+
"low": -0.12411347517730496,
|
| 2250 |
"mean": -0.038534278959810875,
|
| 2251 |
"n": 141,
|
| 2252 |
+
"num_samples": 1000
|
| 2253 |
},
|
| 2254 |
"mean": -0.038534278959810875,
|
| 2255 |
"per_seed": {
|
|
|
|
| 2399 |
"macro_by_task": {
|
| 2400 |
"ci": {
|
| 2401 |
"confidence": 0.95,
|
| 2402 |
+
"high": 0.08220211161387632,
|
| 2403 |
"low": -0.09986928104575163,
|
| 2404 |
"mean": -0.01625942684766214,
|
| 2405 |
"n": 5,
|
| 2406 |
+
"num_samples": 1000
|
| 2407 |
},
|
| 2408 |
"mean": -0.01625942684766214,
|
| 2409 |
"per_task": {
|
|
|
|
| 2418 |
"micro": {
|
| 2419 |
"confidence": 0.95,
|
| 2420 |
"high": 0.024154589371980676,
|
| 2421 |
+
"low": -0.10628019323671498,
|
| 2422 |
"mean": -0.043478260869565216,
|
| 2423 |
"n": 207,
|
| 2424 |
+
"num_samples": 1000
|
| 2425 |
},
|
| 2426 |
"n": 207
|
| 2427 |
},
|
|
|
|
| 2429 |
"macro_by_seed": {
|
| 2430 |
"ci": {
|
| 2431 |
"confidence": 0.95,
|
| 2432 |
+
"high": 0.8302852093747126,
|
| 2433 |
+
"low": 0.5613265780814601,
|
| 2434 |
"mean": 0.6978370489784622,
|
| 2435 |
"n": 141,
|
| 2436 |
+
"num_samples": 1000
|
| 2437 |
},
|
| 2438 |
"mean": 0.6978370489784622,
|
| 2439 |
"per_seed": {
|
|
|
|
| 2584 |
"ci": {
|
| 2585 |
"confidence": 0.95,
|
| 2586 |
"high": 0.886751612873159,
|
| 2587 |
+
"low": 0.2938098484889055,
|
| 2588 |
"mean": 0.5914274646835725,
|
| 2589 |
"n": 5,
|
| 2590 |
+
"num_samples": 1000
|
| 2591 |
},
|
| 2592 |
"mean": 0.5914274646835725,
|
| 2593 |
"per_task": {
|
|
|
|
| 2601 |
"metric": "selected_utility_at_8",
|
| 2602 |
"micro": {
|
| 2603 |
"confidence": 0.95,
|
| 2604 |
+
"high": 0.7017283821891844,
|
| 2605 |
+
"low": 0.48462916244437765,
|
| 2606 |
"mean": 0.5875041471669958,
|
| 2607 |
"n": 207,
|
| 2608 |
+
"num_samples": 1000
|
| 2609 |
},
|
| 2610 |
"n": 207
|
| 2611 |
},
|
|
|
|
| 2613 |
"macro_by_seed": {
|
| 2614 |
"ci": {
|
| 2615 |
"confidence": 0.95,
|
| 2616 |
+
"high": 0.01420507583930653,
|
| 2617 |
+
"low": -0.29225106954879526,
|
| 2618 |
"mean": -0.14455097152575763,
|
| 2619 |
"n": 141,
|
| 2620 |
+
"num_samples": 1000
|
| 2621 |
},
|
| 2622 |
"mean": -0.14455097152575763,
|
| 2623 |
"per_seed": {
|
|
|
|
| 2767 |
"macro_by_task": {
|
| 2768 |
"ci": {
|
| 2769 |
"confidence": 0.95,
|
| 2770 |
+
"high": 0.10282889067617237,
|
| 2771 |
"low": -0.2553936764363997,
|
| 2772 |
"mean": -0.09776939829294695,
|
| 2773 |
"n": 5,
|
| 2774 |
+
"num_samples": 1000
|
| 2775 |
},
|
| 2776 |
"mean": -0.09776939829294695,
|
| 2777 |
"per_task": {
|
|
|
|
| 2785 |
"metric": "selected_utility_gain_over_base_at_8",
|
| 2786 |
"micro": {
|
| 2787 |
"confidence": 0.95,
|
| 2788 |
+
"high": -0.023960533527688885,
|
| 2789 |
"low": -0.26539749033037113,
|
| 2790 |
"mean": -0.14829927152455763,
|
| 2791 |
"n": 207,
|
| 2792 |
+
"num_samples": 1000
|
| 2793 |
},
|
| 2794 |
"n": 207
|
| 2795 |
},
|
|
|
|
| 2797 |
"macro_by_seed": {
|
| 2798 |
"ci": {
|
| 2799 |
"confidence": 0.95,
|
| 2800 |
+
"high": 0.5194647902708409,
|
| 2801 |
+
"low": 0.2999603762294949,
|
| 2802 |
"mean": 0.40561739670197494,
|
| 2803 |
"n": 141,
|
| 2804 |
+
"num_samples": 1000
|
| 2805 |
},
|
| 2806 |
"mean": 0.40561739670197494,
|
| 2807 |
"per_seed": {
|
|
|
|
| 2951 |
"macro_by_task": {
|
| 2952 |
"ci": {
|
| 2953 |
"confidence": 0.95,
|
| 2954 |
+
"high": 0.5319262530406317,
|
| 2955 |
+
"low": 0.19367229081819662,
|
| 2956 |
"mean": 0.3798908010369334,
|
| 2957 |
"n": 5,
|
| 2958 |
+
"num_samples": 1000
|
| 2959 |
},
|
| 2960 |
"mean": 0.3798908010369334,
|
| 2961 |
"per_task": {
|
|
|
|
| 2969 |
"metric": "selector_regret_at_8",
|
| 2970 |
"micro": {
|
| 2971 |
"confidence": 0.95,
|
| 2972 |
+
"high": 0.3721372773162211,
|
| 2973 |
+
"low": 0.21313571063115527,
|
| 2974 |
"mean": 0.293583450403069,
|
| 2975 |
"n": 207,
|
| 2976 |
+
"num_samples": 1000
|
| 2977 |
},
|
| 2978 |
"n": 207
|
| 2979 |
},
|
|
|
|
| 2981 |
"macro_by_seed": {
|
| 2982 |
"ci": {
|
| 2983 |
"confidence": 0.95,
|
| 2984 |
+
"high": 0.2553191489361702,
|
| 2985 |
"low": 0.1276595744680851,
|
| 2986 |
"mean": 0.19148936170212766,
|
| 2987 |
"n": 141,
|
| 2988 |
+
"num_samples": 1000
|
| 2989 |
},
|
| 2990 |
"mean": 0.19148936170212766,
|
| 2991 |
"per_seed": {
|
|
|
|
| 3135 |
"macro_by_task": {
|
| 3136 |
"ci": {
|
| 3137 |
"confidence": 0.95,
|
| 3138 |
+
"high": 0.26170940170940166,
|
| 3139 |
+
"low": 0.05399698340874811,
|
| 3140 |
"mean": 0.15896430367018605,
|
| 3141 |
"n": 5,
|
| 3142 |
+
"num_samples": 1000
|
| 3143 |
},
|
| 3144 |
"mean": 0.15896430367018605,
|
| 3145 |
"per_task": {
|
|
|
|
| 3153 |
"metric": "success_selector_gap_at_8",
|
| 3154 |
"micro": {
|
| 3155 |
"confidence": 0.95,
|
| 3156 |
+
"high": 0.18357487922705315,
|
| 3157 |
+
"low": 0.09178743961352658,
|
| 3158 |
"mean": 0.13526570048309178,
|
| 3159 |
"n": 207,
|
| 3160 |
+
"num_samples": 1000
|
| 3161 |
},
|
| 3162 |
"n": 207
|
| 3163 |
},
|
|
|
|
| 3165 |
"macro_by_seed": {
|
| 3166 |
"ci": {
|
| 3167 |
"confidence": 0.95,
|
| 3168 |
+
"high": 0.43794326241134757,
|
| 3169 |
+
"low": 0.29089834515366436,
|
| 3170 |
"mean": 0.3619385342789597,
|
| 3171 |
"n": 141,
|
| 3172 |
+
"num_samples": 1000
|
| 3173 |
},
|
| 3174 |
"mean": 0.3619385342789597,
|
| 3175 |
"per_seed": {
|
|
|
|
| 3323 |
"low": 0.24550025138260434,
|
| 3324 |
"mean": 0.48667672197083967,
|
| 3325 |
"n": 5,
|
| 3326 |
+
"num_samples": 1000
|
| 3327 |
},
|
| 3328 |
"mean": 0.48667672197083967,
|
| 3329 |
"per_task": {
|
|
|
|
| 3337 |
"metric": "success_support_gap_at_8",
|
| 3338 |
"micro": {
|
| 3339 |
"confidence": 0.95,
|
| 3340 |
+
"high": 0.357487922705314,
|
| 3341 |
"low": 0.2318840579710145,
|
| 3342 |
"mean": 0.2946859903381642,
|
| 3343 |
"n": 207,
|
| 3344 |
+
"num_samples": 1000
|
| 3345 |
},
|
| 3346 |
"n": 207
|
| 3347 |
},
|
|
|
|
| 3349 |
"macro_by_seed": {
|
| 3350 |
"ci": {
|
| 3351 |
"confidence": 0.95,
|
| 3352 |
+
"high": 0.6296690307328605,
|
| 3353 |
+
"low": 0.47482269503546104,
|
| 3354 |
"mean": 0.5534278959810874,
|
| 3355 |
"n": 141,
|
| 3356 |
+
"num_samples": 1000
|
| 3357 |
},
|
| 3358 |
"mean": 0.5534278959810874,
|
| 3359 |
"per_seed": {
|
|
|
|
| 3503 |
"macro_by_task": {
|
| 3504 |
"ci": {
|
| 3505 |
"confidence": 0.95,
|
| 3506 |
+
"high": 0.8844444444444445,
|
| 3507 |
+
"low": 0.3567521367521368,
|
| 3508 |
"mean": 0.6456410256410257,
|
| 3509 |
"n": 5,
|
| 3510 |
+
"num_samples": 1000
|
| 3511 |
},
|
| 3512 |
"mean": 0.6456410256410257,
|
| 3513 |
"per_task": {
|
|
|
|
| 3525 |
"low": 0.357487922705314,
|
| 3526 |
"mean": 0.42995169082125606,
|
| 3527 |
"n": 207,
|
| 3528 |
+
"num_samples": 1000
|
| 3529 |
},
|
| 3530 |
"n": 207
|
| 3531 |
},
|
|
|
|
| 3533 |
"macro_by_seed": {
|
| 3534 |
"ci": {
|
| 3535 |
"confidence": 0.95,
|
| 3536 |
+
"high": 0.7276497844564805,
|
| 3537 |
+
"low": 0.49026043312640605,
|
| 3538 |
"mean": 0.6062044484887313,
|
| 3539 |
"n": 141,
|
| 3540 |
+
"num_samples": 1000
|
| 3541 |
},
|
| 3542 |
"mean": 0.6062044484887313,
|
| 3543 |
"per_seed": {
|
|
|
|
| 3691 |
"low": 0.4585766474838634,
|
| 3692 |
"mean": 0.788961660718036,
|
| 3693 |
"n": 5,
|
| 3694 |
+
"num_samples": 1000
|
| 3695 |
},
|
| 3696 |
"mean": 0.788961660718036,
|
| 3697 |
"per_task": {
|
|
|
|
| 3705 |
"metric": "support_gap_at_8",
|
| 3706 |
"micro": {
|
| 3707 |
"confidence": 0.95,
|
| 3708 |
+
"high": 0.6340587634713039,
|
| 3709 |
"low": 0.4234259423213799,
|
| 3710 |
"mean": 0.5282977284379371,
|
| 3711 |
"n": 207,
|
| 3712 |
+
"num_samples": 1000
|
| 3713 |
},
|
| 3714 |
"n": 207
|
| 3715 |
},
|
|
|
|
| 3717 |
"macro_by_seed": {
|
| 3718 |
"ci": {
|
| 3719 |
"confidence": 0.95,
|
| 3720 |
+
"high": 1.141025779891047,
|
| 3721 |
+
"low": 0.8745643807700053,
|
| 3722 |
"mean": 1.0115917835033712,
|
| 3723 |
"n": 141,
|
| 3724 |
+
"num_samples": 1000
|
| 3725 |
},
|
| 3726 |
"mean": 1.0115917835033712,
|
| 3727 |
"per_seed": {
|
|
|
|
| 3875 |
"low": 0.722751199278172,
|
| 3876 |
"mean": 1.1686045027220224,
|
| 3877 |
"n": 5,
|
| 3878 |
+
"num_samples": 1000
|
| 3879 |
},
|
| 3880 |
"mean": 1.1686045027220224,
|
| 3881 |
"per_task": {
|
|
|
|
| 3889 |
"metric": "total_car_to_hidden_at_8",
|
| 3890 |
"micro": {
|
| 3891 |
"confidence": 0.95,
|
| 3892 |
+
"high": 0.9380562976314821,
|
| 3893 |
+
"low": 0.6983931870469688,
|
| 3894 |
"mean": 0.8212702652815715,
|
| 3895 |
"n": 207,
|
| 3896 |
+
"num_samples": 1000
|
| 3897 |
},
|
| 3898 |
"n": 207
|
| 3899 |
}
|
workspace/runs/ctt_val_rollout_comparison/report.md
CHANGED
|
@@ -7,26 +7,26 @@ K: `8`
|
|
| 7 |
| Metric | N | Micro mean | 95% CI |
|
| 8 |
| --- | ---: | ---: | ---: |
|
| 9 |
| base_success | 207 | 0.2850 | [0.2271, 0.3478] |
|
| 10 |
-
| base_utility | 207 | 0.7358 | [0.
|
| 11 |
-
| branch_car_at_8 | 207 | 0.2936 | [0.
|
| 12 |
-
| hidden_chart_oracle_success_at_8 | 207 | 0.6667 | [0.
|
| 13 |
-
| hidden_chart_oracle_utility_at_8 | 207 | 1.4000 | [1.
|
| 14 |
-
| outcome_ptr_at_8 | 207 | 0.4589 | [0.3913, 0.
|
| 15 |
-
| pairwise_causal_calibration_ece | 207 | 0.1618 | [0.
|
| 16 |
-
| proposal_oracle_success_at_8 | 207 | 0.3768 | [0.3092, 0.
|
| 17 |
-
| proposal_oracle_success_gain_over_base_at_8 | 207 | 0.0918 | [0.
|
| 18 |
-
| proposal_oracle_utility_at_8 | 207 | 0.8811 | [0.
|
| 19 |
-
| proposal_oracle_utility_gain_over_base_at_8 | 207 | 0.1453 | [0.
|
| 20 |
| selected_success_at_8 | 207 | 0.2415 | [0.1884, 0.3043] |
|
| 21 |
-
| selected_success_gain_over_base_at_8 | 207 | -0.0435 | [-0.
|
| 22 |
-
| selected_utility_at_8 | 207 | 0.5875 | [0.
|
| 23 |
-
| selected_utility_gain_over_base_at_8 | 207 | -0.1483 | [-0.2654, -0.
|
| 24 |
-
| selector_regret_at_8 | 207 | 0.2936 | [0.
|
| 25 |
-
| success_selector_gap_at_8 | 207 | 0.1353 | [0.
|
| 26 |
-
| success_support_gap_at_8 | 207 | 0.2947 | [0.2319, 0.
|
| 27 |
| success_total_car_to_hidden_at_8 | 207 | 0.4300 | [0.3575, 0.4976] |
|
| 28 |
-
| support_gap_at_8 | 207 | 0.5283 | [0.4234, 0.
|
| 29 |
-
| total_car_to_hidden_at_8 | 207 | 0.8213 | [0.
|
| 30 |
|
| 31 |
| Success/Utility | Mean |
|
| 32 |
| --- | ---: |
|
|
@@ -42,6 +42,15 @@ K: `8`
|
|
| 42 |
| selected_utility_mean | 0.5875 |
|
| 43 |
| proposal_oracle_utility_mean | 0.8811 |
|
| 44 |
| hidden_chart_oracle_utility_mean | 1.4000 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
|
| 46 |
These are measured generated-candidate rollouts, not PPTC proxies.
|
| 47 |
|
|
|
|
| 7 |
| Metric | N | Micro mean | 95% CI |
|
| 8 |
| --- | ---: | ---: | ---: |
|
| 9 |
| base_success | 207 | 0.2850 | [0.2271, 0.3478] |
|
| 10 |
+
| base_utility | 207 | 0.7358 | [0.6256, 0.8512] |
|
| 11 |
+
| branch_car_at_8 | 207 | 0.2936 | [0.2131, 0.3721] |
|
| 12 |
+
| hidden_chart_oracle_success_at_8 | 207 | 0.6667 | [0.5990, 0.7246] |
|
| 13 |
+
| hidden_chart_oracle_utility_at_8 | 207 | 1.4000 | [1.2800, 1.5063] |
|
| 14 |
+
| outcome_ptr_at_8 | 207 | 0.4589 | [0.3913, 0.5266] |
|
| 15 |
+
| pairwise_causal_calibration_ece | 207 | 0.1618 | [0.1463, 0.1771] |
|
| 16 |
+
| proposal_oracle_success_at_8 | 207 | 0.3768 | [0.3092, 0.4396] |
|
| 17 |
+
| proposal_oracle_success_gain_over_base_at_8 | 207 | 0.0918 | [0.0242, 0.1594] |
|
| 18 |
+
| proposal_oracle_utility_at_8 | 207 | 0.8811 | [0.7614, 1.0027] |
|
| 19 |
+
| proposal_oracle_utility_gain_over_base_at_8 | 207 | 0.1453 | [0.0309, 0.2609] |
|
| 20 |
| selected_success_at_8 | 207 | 0.2415 | [0.1884, 0.3043] |
|
| 21 |
+
| selected_success_gain_over_base_at_8 | 207 | -0.0435 | [-0.1063, 0.0242] |
|
| 22 |
+
| selected_utility_at_8 | 207 | 0.5875 | [0.4846, 0.7017] |
|
| 23 |
+
| selected_utility_gain_over_base_at_8 | 207 | -0.1483 | [-0.2654, -0.0240] |
|
| 24 |
+
| selector_regret_at_8 | 207 | 0.2936 | [0.2131, 0.3721] |
|
| 25 |
+
| success_selector_gap_at_8 | 207 | 0.1353 | [0.0918, 0.1836] |
|
| 26 |
+
| success_support_gap_at_8 | 207 | 0.2947 | [0.2319, 0.3575] |
|
| 27 |
| success_total_car_to_hidden_at_8 | 207 | 0.4300 | [0.3575, 0.4976] |
|
| 28 |
+
| support_gap_at_8 | 207 | 0.5283 | [0.4234, 0.6341] |
|
| 29 |
+
| total_car_to_hidden_at_8 | 207 | 0.8213 | [0.6984, 0.9381] |
|
| 30 |
|
| 31 |
| Success/Utility | Mean |
|
| 32 |
| --- | ---: |
|
|
|
|
| 42 |
| selected_utility_mean | 0.5875 |
|
| 43 |
| proposal_oracle_utility_mean | 0.8811 |
|
| 44 |
| hidden_chart_oracle_utility_mean | 1.4000 |
|
| 45 |
+
| generated_safety_label_coverage | n/a |
|
| 46 |
+
| generated_unsafe_rate_known | n/a |
|
| 47 |
+
| any_generated_unsafe_known | n/a |
|
| 48 |
+
| selected_safety_label_known_rate | n/a |
|
| 49 |
+
| selected_unsafe_rate_known | n/a |
|
| 50 |
+
| proposal_oracle_safety_label_known_rate | n/a |
|
| 51 |
+
| proposal_oracle_unsafe_rate_known | n/a |
|
| 52 |
+
| base_safety_label_known_rate | n/a |
|
| 53 |
+
| base_unsafe_rate_known | n/a |
|
| 54 |
|
| 55 |
These are measured generated-candidate rollouts, not PPTC proxies.
|
| 56 |
|
workspace/runs/ctt_val_rollout_comparison/table.tex
CHANGED
|
@@ -4,25 +4,25 @@
|
|
| 4 |
Metric & N & Micro mean & CI low & CI high \\
|
| 5 |
\midrule
|
| 6 |
base\_success & 207 & 0.2850 & 0.2271 & 0.3478 \\
|
| 7 |
-
base\_utility & 207 & 0.7358 & 0.
|
| 8 |
-
branch\_car\_at\_8 & 207 & 0.2936 & 0.
|
| 9 |
-
hidden\_chart\_oracle\_success\_at\_8 & 207 & 0.6667 & 0.
|
| 10 |
-
hidden\_chart\_oracle\_utility\_at\_8 & 207 & 1.4000 & 1.
|
| 11 |
-
outcome\_ptr\_at\_8 & 207 & 0.4589 & 0.3913 & 0.
|
| 12 |
-
pairwise\_causal\_calibration\_ece & 207 & 0.1618 & 0.
|
| 13 |
-
proposal\_oracle\_success\_at\_8 & 207 & 0.3768 & 0.3092 & 0.
|
| 14 |
-
proposal\_oracle\_success\_gain\_over\_base\_at\_8 & 207 & 0.0918 & 0.
|
| 15 |
-
proposal\_oracle\_utility\_at\_8 & 207 & 0.8811 & 0.
|
| 16 |
-
proposal\_oracle\_utility\_gain\_over\_base\_at\_8 & 207 & 0.1453 & 0.
|
| 17 |
selected\_success\_at\_8 & 207 & 0.2415 & 0.1884 & 0.3043 \\
|
| 18 |
-
selected\_success\_gain\_over\_base\_at\_8 & 207 & -0.0435 & -0.
|
| 19 |
-
selected\_utility\_at\_8 & 207 & 0.5875 & 0.
|
| 20 |
-
selected\_utility\_gain\_over\_base\_at\_8 & 207 & -0.1483 & -0.2654 & -0.
|
| 21 |
-
selector\_regret\_at\_8 & 207 & 0.2936 & 0.
|
| 22 |
-
success\_selector\_gap\_at\_8 & 207 & 0.1353 & 0.
|
| 23 |
-
success\_support\_gap\_at\_8 & 207 & 0.2947 & 0.2319 & 0.
|
| 24 |
success\_total\_car\_to\_hidden\_at\_8 & 207 & 0.4300 & 0.3575 & 0.4976 \\
|
| 25 |
-
support\_gap\_at\_8 & 207 & 0.5283 & 0.4234 & 0.
|
| 26 |
-
total\_car\_to\_hidden\_at\_8 & 207 & 0.8213 & 0.
|
| 27 |
\bottomrule
|
| 28 |
\end{tabular}
|
|
|
|
| 4 |
Metric & N & Micro mean & CI low & CI high \\
|
| 5 |
\midrule
|
| 6 |
base\_success & 207 & 0.2850 & 0.2271 & 0.3478 \\
|
| 7 |
+
base\_utility & 207 & 0.7358 & 0.6256 & 0.8512 \\
|
| 8 |
+
branch\_car\_at\_8 & 207 & 0.2936 & 0.2131 & 0.3721 \\
|
| 9 |
+
hidden\_chart\_oracle\_success\_at\_8 & 207 & 0.6667 & 0.5990 & 0.7246 \\
|
| 10 |
+
hidden\_chart\_oracle\_utility\_at\_8 & 207 & 1.4000 & 1.2800 & 1.5063 \\
|
| 11 |
+
outcome\_ptr\_at\_8 & 207 & 0.4589 & 0.3913 & 0.5266 \\
|
| 12 |
+
pairwise\_causal\_calibration\_ece & 207 & 0.1618 & 0.1463 & 0.1771 \\
|
| 13 |
+
proposal\_oracle\_success\_at\_8 & 207 & 0.3768 & 0.3092 & 0.4396 \\
|
| 14 |
+
proposal\_oracle\_success\_gain\_over\_base\_at\_8 & 207 & 0.0918 & 0.0242 & 0.1594 \\
|
| 15 |
+
proposal\_oracle\_utility\_at\_8 & 207 & 0.8811 & 0.7614 & 1.0027 \\
|
| 16 |
+
proposal\_oracle\_utility\_gain\_over\_base\_at\_8 & 207 & 0.1453 & 0.0309 & 0.2609 \\
|
| 17 |
selected\_success\_at\_8 & 207 & 0.2415 & 0.1884 & 0.3043 \\
|
| 18 |
+
selected\_success\_gain\_over\_base\_at\_8 & 207 & -0.0435 & -0.1063 & 0.0242 \\
|
| 19 |
+
selected\_utility\_at\_8 & 207 & 0.5875 & 0.4846 & 0.7017 \\
|
| 20 |
+
selected\_utility\_gain\_over\_base\_at\_8 & 207 & -0.1483 & -0.2654 & -0.0240 \\
|
| 21 |
+
selector\_regret\_at\_8 & 207 & 0.2936 & 0.2131 & 0.3721 \\
|
| 22 |
+
success\_selector\_gap\_at\_8 & 207 & 0.1353 & 0.0918 & 0.1836 \\
|
| 23 |
+
success\_support\_gap\_at\_8 & 207 & 0.2947 & 0.2319 & 0.3575 \\
|
| 24 |
success\_total\_car\_to\_hidden\_at\_8 & 207 & 0.4300 & 0.3575 & 0.4976 \\
|
| 25 |
+
support\_gap\_at\_8 & 207 & 0.5283 & 0.4234 & 0.6341 \\
|
| 26 |
+
total\_car\_to\_hidden\_at\_8 & 207 & 0.8213 & 0.6984 & 0.9381 \\
|
| 27 |
\bottomrule
|
| 28 |
\end{tabular}
|
workspace/runs/leakage_audit_rgb_refs_object/index_report.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"indexes": [
|
| 3 |
+
{
|
| 4 |
+
"audience": "evaluator_only",
|
| 5 |
+
"include_outcomes": true,
|
| 6 |
+
"num_groups": 410,
|
| 7 |
+
"num_rows": 6560,
|
| 8 |
+
"path": "data/cil_charts_rgb_refs/test/index.json",
|
| 9 |
+
"retrieval_index_allowed": false,
|
| 10 |
+
"split": "test"
|
| 11 |
+
},
|
| 12 |
+
{
|
| 13 |
+
"audience": "train_retrieval",
|
| 14 |
+
"include_outcomes": true,
|
| 15 |
+
"num_groups": 2044,
|
| 16 |
+
"num_rows": 32704,
|
| 17 |
+
"path": "data/cil_charts_rgb_refs/train/index.json",
|
| 18 |
+
"retrieval_index_allowed": true,
|
| 19 |
+
"split": "train"
|
| 20 |
+
},
|
| 21 |
+
{
|
| 22 |
+
"audience": "evaluator_only",
|
| 23 |
+
"include_outcomes": true,
|
| 24 |
+
"num_groups": 419,
|
| 25 |
+
"num_rows": 6704,
|
| 26 |
+
"path": "data/cil_charts_rgb_refs/val/index.json",
|
| 27 |
+
"retrieval_index_allowed": false,
|
| 28 |
+
"split": "val"
|
| 29 |
+
}
|
| 30 |
+
],
|
| 31 |
+
"num_indexes": 3,
|
| 32 |
+
"status": "pass",
|
| 33 |
+
"violations": [],
|
| 34 |
+
"warnings": []
|
| 35 |
+
}
|
workspace/runs/leakage_audit_rgb_refs_object/report.json
ADDED
|
@@ -0,0 +1,159 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"chart_root": "data/cil_charts_rgb_refs",
|
| 3 |
+
"indexes": {
|
| 4 |
+
"test": {
|
| 5 |
+
"audience": "evaluator_only",
|
| 6 |
+
"content_hash": "0c38d73dc9bc4ab756e12006671b5d12f1a23358b281e54b3f6a9ae3cbf6e9af",
|
| 7 |
+
"include_outcomes": true,
|
| 8 |
+
"num_groups_exported": 410,
|
| 9 |
+
"num_rows": 6560,
|
| 10 |
+
"path": "data/cil_charts_rgb_refs/test/index.json",
|
| 11 |
+
"retrieval_index_allowed": false,
|
| 12 |
+
"split_hash": "84151df9e88c219b76751677558cc5bbbc19082bf358cea7330158e9e403e25f"
|
| 13 |
+
},
|
| 14 |
+
"train": {
|
| 15 |
+
"audience": "train_retrieval",
|
| 16 |
+
"content_hash": "67f7b4a692f7c7e71da24378dc71e3399c5d35afc14c494869eba6087b765c42",
|
| 17 |
+
"include_outcomes": true,
|
| 18 |
+
"num_groups_exported": 2044,
|
| 19 |
+
"num_rows": 32704,
|
| 20 |
+
"path": "data/cil_charts_rgb_refs/train/index.json",
|
| 21 |
+
"retrieval_index_allowed": true,
|
| 22 |
+
"split_hash": "867a94a1c67686c082765671a7bf8d74a9d633085e1b2619c83197c60bf58372"
|
| 23 |
+
},
|
| 24 |
+
"val": {
|
| 25 |
+
"audience": "evaluator_only",
|
| 26 |
+
"content_hash": "1aacf9de98648fa36bef14e137eb43c7c6973170804650814dae41d09842c708",
|
| 27 |
+
"include_outcomes": true,
|
| 28 |
+
"num_groups_exported": 419,
|
| 29 |
+
"num_rows": 6704,
|
| 30 |
+
"path": "data/cil_charts_rgb_refs/val/index.json",
|
| 31 |
+
"retrieval_index_allowed": false,
|
| 32 |
+
"split_hash": "9d1dc1c10868f76fa080895df0f2730c173f12b6db14932599d55c2b60859381"
|
| 33 |
+
}
|
| 34 |
+
},
|
| 35 |
+
"status": "pass",
|
| 36 |
+
"violations": [],
|
| 37 |
+
"warnings": [
|
| 38 |
+
"runs/chart_feature_audit/metrics.json: missing data_hash",
|
| 39 |
+
"runs/chart_feature_audit/metrics.json: missing split_hash",
|
| 40 |
+
"runs/chart_feature_audit_rgb_refs/metrics.json: missing data_hash",
|
| 41 |
+
"runs/chart_feature_audit_rgb_refs/metrics.json: missing split_hash",
|
| 42 |
+
"runs/ctt_base_context_obs_dominance_train_to_test/metrics.json: missing data_hash",
|
| 43 |
+
"runs/ctt_base_context_obs_dominance_train_to_test/metrics.json: missing split_hash",
|
| 44 |
+
"runs/ctt_base_context_obs_dominance_train_to_test_tau0/metrics.json: missing data_hash",
|
| 45 |
+
"runs/ctt_base_context_obs_dominance_train_to_test_tau0/metrics.json: missing split_hash",
|
| 46 |
+
"runs/ctt_base_context_obs_dominance_val_to_test/metrics.json: missing data_hash",
|
| 47 |
+
"runs/ctt_base_context_obs_dominance_val_to_test/metrics.json: missing split_hash",
|
| 48 |
+
"runs/ctt_base_context_obs_dominance_val_to_test_tau0/metrics.json: missing data_hash",
|
| 49 |
+
"runs/ctt_base_context_obs_dominance_val_to_test_tau0/metrics.json: missing split_hash",
|
| 50 |
+
"runs/ctt_base_context_obs_learned_dominance_context_success_weighted_train_to_test/metrics.json: missing data_hash",
|
| 51 |
+
"runs/ctt_base_context_obs_learned_dominance_context_success_weighted_train_to_test/metrics.json: missing split_hash",
|
| 52 |
+
"runs/ctt_base_context_obs_learned_dominance_context_success_weighted_val_to_test/metrics.json: missing data_hash",
|
| 53 |
+
"runs/ctt_base_context_obs_learned_dominance_context_success_weighted_val_to_test/metrics.json: missing split_hash",
|
| 54 |
+
"runs/ctt_base_context_obs_learned_dominance_context_tangent_train_to_test/metrics.json: missing data_hash",
|
| 55 |
+
"runs/ctt_base_context_obs_learned_dominance_context_tangent_train_to_test/metrics.json: missing split_hash",
|
| 56 |
+
"runs/ctt_base_context_obs_learned_dominance_context_tangent_val_to_test/metrics.json: missing data_hash",
|
| 57 |
+
"runs/ctt_base_context_obs_learned_dominance_context_tangent_val_to_test/metrics.json: missing split_hash",
|
| 58 |
+
"runs/ctt_base_context_obs_learned_dominance_context_train_to_test/metrics.json: missing data_hash",
|
| 59 |
+
"runs/ctt_base_context_obs_learned_dominance_context_train_to_test/metrics.json: missing split_hash",
|
| 60 |
+
"runs/ctt_base_context_obs_learned_dominance_context_val_to_test/metrics.json: missing data_hash",
|
| 61 |
+
"runs/ctt_base_context_obs_learned_dominance_context_val_to_test/metrics.json: missing split_hash",
|
| 62 |
+
"runs/ctt_base_context_obs_learned_dominance_success_train_to_test/metrics.json: missing data_hash",
|
| 63 |
+
"runs/ctt_base_context_obs_learned_dominance_success_train_to_test/metrics.json: missing split_hash",
|
| 64 |
+
"runs/ctt_base_context_obs_learned_dominance_success_val_to_test/metrics.json: missing data_hash",
|
| 65 |
+
"runs/ctt_base_context_obs_learned_dominance_success_val_to_test/metrics.json: missing split_hash",
|
| 66 |
+
"runs/ctt_base_context_obs_learned_dominance_success_weighted_train_to_test/metrics.json: missing data_hash",
|
| 67 |
+
"runs/ctt_base_context_obs_learned_dominance_success_weighted_train_to_test/metrics.json: missing split_hash",
|
| 68 |
+
"runs/ctt_base_context_obs_learned_dominance_success_weighted_val_to_test/metrics.json: missing data_hash",
|
| 69 |
+
"runs/ctt_base_context_obs_learned_dominance_success_weighted_val_to_test/metrics.json: missing split_hash",
|
| 70 |
+
"runs/ctt_base_context_obs_learned_dominance_tangent_val_to_test/metrics.json: missing data_hash",
|
| 71 |
+
"runs/ctt_base_context_obs_learned_dominance_tangent_val_to_test/metrics.json: missing split_hash",
|
| 72 |
+
"runs/ctt_base_context_obs_learned_dominance_train_to_test/metrics.json: missing data_hash",
|
| 73 |
+
"runs/ctt_base_context_obs_learned_dominance_train_to_test/metrics.json: missing split_hash",
|
| 74 |
+
"runs/ctt_base_context_obs_learned_dominance_val_to_test/metrics.json: missing data_hash",
|
| 75 |
+
"runs/ctt_base_context_obs_learned_dominance_val_to_test/metrics.json: missing split_hash",
|
| 76 |
+
"runs/ctt_base_context_obs_nonlinear_dominance_basic_positive_train_to_test/metrics.json: missing data_hash",
|
| 77 |
+
"runs/ctt_base_context_obs_nonlinear_dominance_basic_positive_train_to_test/metrics.json: missing split_hash",
|
| 78 |
+
"runs/ctt_base_context_obs_nonlinear_dominance_context_success_train_to_test/metrics.json: missing data_hash",
|
| 79 |
+
"runs/ctt_base_context_obs_nonlinear_dominance_context_success_train_to_test/metrics.json: missing split_hash",
|
| 80 |
+
"runs/ctt_base_context_obs_nonlinear_dominance_context_tangent_positive_train_to_test/metrics.json: missing data_hash",
|
| 81 |
+
"runs/ctt_base_context_obs_nonlinear_dominance_context_tangent_positive_train_to_test/metrics.json: missing split_hash",
|
| 82 |
+
"runs/ctt_base_context_obs_test_rollout_comparison/metrics.json: missing data_hash",
|
| 83 |
+
"runs/ctt_base_context_obs_test_rollout_comparison/metrics.json: missing split_hash",
|
| 84 |
+
"runs/ctt_base_context_obs_train_cal_rollout_comparison/metrics.json: missing data_hash",
|
| 85 |
+
"runs/ctt_base_context_obs_train_cal_rollout_comparison/metrics.json: missing split_hash",
|
| 86 |
+
"runs/ctt_base_context_obs_val_rollout_comparison/metrics.json: missing data_hash",
|
| 87 |
+
"runs/ctt_base_context_obs_val_rollout_comparison/metrics.json: missing split_hash",
|
| 88 |
+
"runs/ctt_dominance_utility_energy_smoke_val_to_test/metrics.json: missing data_hash",
|
| 89 |
+
"runs/ctt_dominance_utility_energy_smoke_val_to_test/metrics.json: missing split_hash",
|
| 90 |
+
"runs/ctt_dominance_utility_energy_val_to_test_seed0/metrics.json: missing data_hash",
|
| 91 |
+
"runs/ctt_dominance_utility_energy_val_to_test_seed0/metrics.json: missing split_hash",
|
| 92 |
+
"runs/ctt_dominance_utility_energy_val_to_test_seed1/metrics.json: missing data_hash",
|
| 93 |
+
"runs/ctt_dominance_utility_energy_val_to_test_seed1/metrics.json: missing split_hash",
|
| 94 |
+
"runs/ctt_dominance_utility_energy_val_to_test_seed2/metrics.json: missing data_hash",
|
| 95 |
+
"runs/ctt_dominance_utility_energy_val_to_test_seed2/metrics.json: missing split_hash",
|
| 96 |
+
"runs/ctt_dominance_val_to_test/metrics.json: missing data_hash",
|
| 97 |
+
"runs/ctt_dominance_val_to_test/metrics.json: missing split_hash",
|
| 98 |
+
"runs/ctt_dominance_val_to_test_tau0/metrics.json: missing data_hash",
|
| 99 |
+
"runs/ctt_dominance_val_to_test_tau0/metrics.json: missing split_hash",
|
| 100 |
+
"runs/ctt_learned_dominance_context_success_weighted_val_to_test/metrics.json: missing data_hash",
|
| 101 |
+
"runs/ctt_learned_dominance_context_success_weighted_val_to_test/metrics.json: missing split_hash",
|
| 102 |
+
"runs/ctt_learned_dominance_context_tangent_val_to_test/metrics.json: missing data_hash",
|
| 103 |
+
"runs/ctt_learned_dominance_context_tangent_val_to_test/metrics.json: missing split_hash",
|
| 104 |
+
"runs/ctt_learned_dominance_context_val_to_test/metrics.json: missing data_hash",
|
| 105 |
+
"runs/ctt_learned_dominance_context_val_to_test/metrics.json: missing split_hash",
|
| 106 |
+
"runs/ctt_learned_dominance_margin_ext_val_to_test/metrics.json: missing data_hash",
|
| 107 |
+
"runs/ctt_learned_dominance_margin_ext_val_to_test/metrics.json: missing split_hash",
|
| 108 |
+
"runs/ctt_learned_dominance_success_val_to_test/metrics.json: missing data_hash",
|
| 109 |
+
"runs/ctt_learned_dominance_success_val_to_test/metrics.json: missing split_hash",
|
| 110 |
+
"runs/ctt_learned_dominance_success_weighted_val_to_test/metrics.json: missing data_hash",
|
| 111 |
+
"runs/ctt_learned_dominance_success_weighted_val_to_test/metrics.json: missing split_hash",
|
| 112 |
+
"runs/ctt_learned_dominance_tangent_val_to_test/metrics.json: missing data_hash",
|
| 113 |
+
"runs/ctt_learned_dominance_tangent_val_to_test/metrics.json: missing split_hash",
|
| 114 |
+
"runs/ctt_learned_dominance_val_to_test/metrics.json: missing data_hash",
|
| 115 |
+
"runs/ctt_learned_dominance_val_to_test/metrics.json: missing split_hash",
|
| 116 |
+
"runs/ctt_residual_base_context_obs_rollout_test_seed0/metrics.json: missing data_hash",
|
| 117 |
+
"runs/ctt_residual_base_context_obs_rollout_test_seed0/metrics.json: missing split_hash",
|
| 118 |
+
"runs/ctt_residual_base_context_obs_rollout_test_seed1/metrics.json: missing data_hash",
|
| 119 |
+
"runs/ctt_residual_base_context_obs_rollout_test_seed1/metrics.json: missing split_hash",
|
| 120 |
+
"runs/ctt_residual_base_context_obs_rollout_test_seed2/metrics.json: missing data_hash",
|
| 121 |
+
"runs/ctt_residual_base_context_obs_rollout_test_seed2/metrics.json: missing split_hash",
|
| 122 |
+
"runs/ctt_residual_base_context_obs_rollout_train_cal_seed0/metrics.json: missing data_hash",
|
| 123 |
+
"runs/ctt_residual_base_context_obs_rollout_train_cal_seed0/metrics.json: missing split_hash",
|
| 124 |
+
"runs/ctt_residual_base_context_obs_rollout_train_cal_seed1/metrics.json: missing data_hash",
|
| 125 |
+
"runs/ctt_residual_base_context_obs_rollout_train_cal_seed1/metrics.json: missing split_hash",
|
| 126 |
+
"runs/ctt_residual_base_context_obs_rollout_train_cal_seed2/metrics.json: missing data_hash",
|
| 127 |
+
"runs/ctt_residual_base_context_obs_rollout_train_cal_seed2/metrics.json: missing split_hash",
|
| 128 |
+
"runs/ctt_residual_base_context_obs_rollout_val69_seed0/metrics.json: missing data_hash",
|
| 129 |
+
"runs/ctt_residual_base_context_obs_rollout_val69_seed0/metrics.json: missing split_hash",
|
| 130 |
+
"runs/ctt_residual_base_context_obs_rollout_val69_seed1/metrics.json: missing data_hash",
|
| 131 |
+
"runs/ctt_residual_base_context_obs_rollout_val69_seed1/metrics.json: missing split_hash",
|
| 132 |
+
"runs/ctt_residual_base_context_obs_rollout_val69_seed2/metrics.json: missing data_hash",
|
| 133 |
+
"runs/ctt_residual_base_context_obs_rollout_val69_seed2/metrics.json: missing split_hash",
|
| 134 |
+
"runs/ctt_residual_rollout_direct_smoke_seed0_v3/metrics.json: missing data_hash",
|
| 135 |
+
"runs/ctt_residual_rollout_direct_smoke_seed0_v3/metrics.json: missing split_hash",
|
| 136 |
+
"runs/ctt_residual_rollout_test_seed0/metrics.json: missing data_hash",
|
| 137 |
+
"runs/ctt_residual_rollout_test_seed0/metrics.json: missing split_hash",
|
| 138 |
+
"runs/ctt_residual_rollout_test_seed1/metrics.json: missing data_hash",
|
| 139 |
+
"runs/ctt_residual_rollout_test_seed1/metrics.json: missing split_hash",
|
| 140 |
+
"runs/ctt_residual_rollout_test_seed2/metrics.json: missing data_hash",
|
| 141 |
+
"runs/ctt_residual_rollout_test_seed2/metrics.json: missing split_hash",
|
| 142 |
+
"runs/ctt_residual_rollout_val16_seed0/metrics.json: missing data_hash",
|
| 143 |
+
"runs/ctt_residual_rollout_val16_seed0/metrics.json: missing split_hash",
|
| 144 |
+
"runs/ctt_residual_rollout_val69_seed0/metrics.json: missing data_hash",
|
| 145 |
+
"runs/ctt_residual_rollout_val69_seed0/metrics.json: missing split_hash",
|
| 146 |
+
"runs/ctt_residual_rollout_val69_seed1/metrics.json: missing data_hash",
|
| 147 |
+
"runs/ctt_residual_rollout_val69_seed1/metrics.json: missing split_hash",
|
| 148 |
+
"runs/ctt_residual_rollout_val69_seed2/metrics.json: missing data_hash",
|
| 149 |
+
"runs/ctt_residual_rollout_val69_seed2/metrics.json: missing split_hash",
|
| 150 |
+
"runs/ctt_test_rollout_comparison/metrics.json: missing data_hash",
|
| 151 |
+
"runs/ctt_test_rollout_comparison/metrics.json: missing split_hash",
|
| 152 |
+
"runs/ctt_val16_rollout_comparison/metrics.json: missing data_hash",
|
| 153 |
+
"runs/ctt_val16_rollout_comparison/metrics.json: missing split_hash",
|
| 154 |
+
"runs/ctt_val_rollout_comparison/metrics.json: missing data_hash",
|
| 155 |
+
"runs/ctt_val_rollout_comparison/metrics.json: missing split_hash",
|
| 156 |
+
"runs/tangent_reconstruction/metrics.json: missing data_hash",
|
| 157 |
+
"runs/tangent_reconstruction/metrics.json: missing split_hash"
|
| 158 |
+
]
|
| 159 |
+
}
|
workspace/runs/leakage_audit_rgb_refs_object/report.md
ADDED
|
@@ -0,0 +1,136 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# CIL Chart Leakage Audit
|
| 2 |
+
|
| 3 |
+
Status: `pass`
|
| 4 |
+
|
| 5 |
+
| Split | Rows | Charts | Outcomes in DB | Retrieval allowed | Audience |
|
| 6 |
+
| --- | ---: | ---: | --- | --- | --- |
|
| 7 |
+
| test | 6560 | 410 | True | False | evaluator_only |
|
| 8 |
+
| train | 32704 | 2044 | True | True | train_retrieval |
|
| 9 |
+
| val | 6704 | 419 | True | False | evaluator_only |
|
| 10 |
+
|
| 11 |
+
## Violations
|
| 12 |
+
|
| 13 |
+
- None
|
| 14 |
+
|
| 15 |
+
## Warnings
|
| 16 |
+
|
| 17 |
+
- runs/chart_feature_audit/metrics.json: missing data_hash
|
| 18 |
+
- runs/chart_feature_audit/metrics.json: missing split_hash
|
| 19 |
+
- runs/chart_feature_audit_rgb_refs/metrics.json: missing data_hash
|
| 20 |
+
- runs/chart_feature_audit_rgb_refs/metrics.json: missing split_hash
|
| 21 |
+
- runs/ctt_base_context_obs_dominance_train_to_test/metrics.json: missing data_hash
|
| 22 |
+
- runs/ctt_base_context_obs_dominance_train_to_test/metrics.json: missing split_hash
|
| 23 |
+
- runs/ctt_base_context_obs_dominance_train_to_test_tau0/metrics.json: missing data_hash
|
| 24 |
+
- runs/ctt_base_context_obs_dominance_train_to_test_tau0/metrics.json: missing split_hash
|
| 25 |
+
- runs/ctt_base_context_obs_dominance_val_to_test/metrics.json: missing data_hash
|
| 26 |
+
- runs/ctt_base_context_obs_dominance_val_to_test/metrics.json: missing split_hash
|
| 27 |
+
- runs/ctt_base_context_obs_dominance_val_to_test_tau0/metrics.json: missing data_hash
|
| 28 |
+
- runs/ctt_base_context_obs_dominance_val_to_test_tau0/metrics.json: missing split_hash
|
| 29 |
+
- runs/ctt_base_context_obs_learned_dominance_context_success_weighted_train_to_test/metrics.json: missing data_hash
|
| 30 |
+
- runs/ctt_base_context_obs_learned_dominance_context_success_weighted_train_to_test/metrics.json: missing split_hash
|
| 31 |
+
- runs/ctt_base_context_obs_learned_dominance_context_success_weighted_val_to_test/metrics.json: missing data_hash
|
| 32 |
+
- runs/ctt_base_context_obs_learned_dominance_context_success_weighted_val_to_test/metrics.json: missing split_hash
|
| 33 |
+
- runs/ctt_base_context_obs_learned_dominance_context_tangent_train_to_test/metrics.json: missing data_hash
|
| 34 |
+
- runs/ctt_base_context_obs_learned_dominance_context_tangent_train_to_test/metrics.json: missing split_hash
|
| 35 |
+
- runs/ctt_base_context_obs_learned_dominance_context_tangent_val_to_test/metrics.json: missing data_hash
|
| 36 |
+
- runs/ctt_base_context_obs_learned_dominance_context_tangent_val_to_test/metrics.json: missing split_hash
|
| 37 |
+
- runs/ctt_base_context_obs_learned_dominance_context_train_to_test/metrics.json: missing data_hash
|
| 38 |
+
- runs/ctt_base_context_obs_learned_dominance_context_train_to_test/metrics.json: missing split_hash
|
| 39 |
+
- runs/ctt_base_context_obs_learned_dominance_context_val_to_test/metrics.json: missing data_hash
|
| 40 |
+
- runs/ctt_base_context_obs_learned_dominance_context_val_to_test/metrics.json: missing split_hash
|
| 41 |
+
- runs/ctt_base_context_obs_learned_dominance_success_train_to_test/metrics.json: missing data_hash
|
| 42 |
+
- runs/ctt_base_context_obs_learned_dominance_success_train_to_test/metrics.json: missing split_hash
|
| 43 |
+
- runs/ctt_base_context_obs_learned_dominance_success_val_to_test/metrics.json: missing data_hash
|
| 44 |
+
- runs/ctt_base_context_obs_learned_dominance_success_val_to_test/metrics.json: missing split_hash
|
| 45 |
+
- runs/ctt_base_context_obs_learned_dominance_success_weighted_train_to_test/metrics.json: missing data_hash
|
| 46 |
+
- runs/ctt_base_context_obs_learned_dominance_success_weighted_train_to_test/metrics.json: missing split_hash
|
| 47 |
+
- runs/ctt_base_context_obs_learned_dominance_success_weighted_val_to_test/metrics.json: missing data_hash
|
| 48 |
+
- runs/ctt_base_context_obs_learned_dominance_success_weighted_val_to_test/metrics.json: missing split_hash
|
| 49 |
+
- runs/ctt_base_context_obs_learned_dominance_tangent_val_to_test/metrics.json: missing data_hash
|
| 50 |
+
- runs/ctt_base_context_obs_learned_dominance_tangent_val_to_test/metrics.json: missing split_hash
|
| 51 |
+
- runs/ctt_base_context_obs_learned_dominance_train_to_test/metrics.json: missing data_hash
|
| 52 |
+
- runs/ctt_base_context_obs_learned_dominance_train_to_test/metrics.json: missing split_hash
|
| 53 |
+
- runs/ctt_base_context_obs_learned_dominance_val_to_test/metrics.json: missing data_hash
|
| 54 |
+
- runs/ctt_base_context_obs_learned_dominance_val_to_test/metrics.json: missing split_hash
|
| 55 |
+
- runs/ctt_base_context_obs_nonlinear_dominance_basic_positive_train_to_test/metrics.json: missing data_hash
|
| 56 |
+
- runs/ctt_base_context_obs_nonlinear_dominance_basic_positive_train_to_test/metrics.json: missing split_hash
|
| 57 |
+
- runs/ctt_base_context_obs_nonlinear_dominance_context_success_train_to_test/metrics.json: missing data_hash
|
| 58 |
+
- runs/ctt_base_context_obs_nonlinear_dominance_context_success_train_to_test/metrics.json: missing split_hash
|
| 59 |
+
- runs/ctt_base_context_obs_nonlinear_dominance_context_tangent_positive_train_to_test/metrics.json: missing data_hash
|
| 60 |
+
- runs/ctt_base_context_obs_nonlinear_dominance_context_tangent_positive_train_to_test/metrics.json: missing split_hash
|
| 61 |
+
- runs/ctt_base_context_obs_test_rollout_comparison/metrics.json: missing data_hash
|
| 62 |
+
- runs/ctt_base_context_obs_test_rollout_comparison/metrics.json: missing split_hash
|
| 63 |
+
- runs/ctt_base_context_obs_train_cal_rollout_comparison/metrics.json: missing data_hash
|
| 64 |
+
- runs/ctt_base_context_obs_train_cal_rollout_comparison/metrics.json: missing split_hash
|
| 65 |
+
- runs/ctt_base_context_obs_val_rollout_comparison/metrics.json: missing data_hash
|
| 66 |
+
- runs/ctt_base_context_obs_val_rollout_comparison/metrics.json: missing split_hash
|
| 67 |
+
- runs/ctt_dominance_utility_energy_smoke_val_to_test/metrics.json: missing data_hash
|
| 68 |
+
- runs/ctt_dominance_utility_energy_smoke_val_to_test/metrics.json: missing split_hash
|
| 69 |
+
- runs/ctt_dominance_utility_energy_val_to_test_seed0/metrics.json: missing data_hash
|
| 70 |
+
- runs/ctt_dominance_utility_energy_val_to_test_seed0/metrics.json: missing split_hash
|
| 71 |
+
- runs/ctt_dominance_utility_energy_val_to_test_seed1/metrics.json: missing data_hash
|
| 72 |
+
- runs/ctt_dominance_utility_energy_val_to_test_seed1/metrics.json: missing split_hash
|
| 73 |
+
- runs/ctt_dominance_utility_energy_val_to_test_seed2/metrics.json: missing data_hash
|
| 74 |
+
- runs/ctt_dominance_utility_energy_val_to_test_seed2/metrics.json: missing split_hash
|
| 75 |
+
- runs/ctt_dominance_val_to_test/metrics.json: missing data_hash
|
| 76 |
+
- runs/ctt_dominance_val_to_test/metrics.json: missing split_hash
|
| 77 |
+
- runs/ctt_dominance_val_to_test_tau0/metrics.json: missing data_hash
|
| 78 |
+
- runs/ctt_dominance_val_to_test_tau0/metrics.json: missing split_hash
|
| 79 |
+
- runs/ctt_learned_dominance_context_success_weighted_val_to_test/metrics.json: missing data_hash
|
| 80 |
+
- runs/ctt_learned_dominance_context_success_weighted_val_to_test/metrics.json: missing split_hash
|
| 81 |
+
- runs/ctt_learned_dominance_context_tangent_val_to_test/metrics.json: missing data_hash
|
| 82 |
+
- runs/ctt_learned_dominance_context_tangent_val_to_test/metrics.json: missing split_hash
|
| 83 |
+
- runs/ctt_learned_dominance_context_val_to_test/metrics.json: missing data_hash
|
| 84 |
+
- runs/ctt_learned_dominance_context_val_to_test/metrics.json: missing split_hash
|
| 85 |
+
- runs/ctt_learned_dominance_margin_ext_val_to_test/metrics.json: missing data_hash
|
| 86 |
+
- runs/ctt_learned_dominance_margin_ext_val_to_test/metrics.json: missing split_hash
|
| 87 |
+
- runs/ctt_learned_dominance_success_val_to_test/metrics.json: missing data_hash
|
| 88 |
+
- runs/ctt_learned_dominance_success_val_to_test/metrics.json: missing split_hash
|
| 89 |
+
- runs/ctt_learned_dominance_success_weighted_val_to_test/metrics.json: missing data_hash
|
| 90 |
+
- runs/ctt_learned_dominance_success_weighted_val_to_test/metrics.json: missing split_hash
|
| 91 |
+
- runs/ctt_learned_dominance_tangent_val_to_test/metrics.json: missing data_hash
|
| 92 |
+
- runs/ctt_learned_dominance_tangent_val_to_test/metrics.json: missing split_hash
|
| 93 |
+
- runs/ctt_learned_dominance_val_to_test/metrics.json: missing data_hash
|
| 94 |
+
- runs/ctt_learned_dominance_val_to_test/metrics.json: missing split_hash
|
| 95 |
+
- runs/ctt_residual_base_context_obs_rollout_test_seed0/metrics.json: missing data_hash
|
| 96 |
+
- runs/ctt_residual_base_context_obs_rollout_test_seed0/metrics.json: missing split_hash
|
| 97 |
+
- runs/ctt_residual_base_context_obs_rollout_test_seed1/metrics.json: missing data_hash
|
| 98 |
+
- runs/ctt_residual_base_context_obs_rollout_test_seed1/metrics.json: missing split_hash
|
| 99 |
+
- runs/ctt_residual_base_context_obs_rollout_test_seed2/metrics.json: missing data_hash
|
| 100 |
+
- runs/ctt_residual_base_context_obs_rollout_test_seed2/metrics.json: missing split_hash
|
| 101 |
+
- runs/ctt_residual_base_context_obs_rollout_train_cal_seed0/metrics.json: missing data_hash
|
| 102 |
+
- runs/ctt_residual_base_context_obs_rollout_train_cal_seed0/metrics.json: missing split_hash
|
| 103 |
+
- runs/ctt_residual_base_context_obs_rollout_train_cal_seed1/metrics.json: missing data_hash
|
| 104 |
+
- runs/ctt_residual_base_context_obs_rollout_train_cal_seed1/metrics.json: missing split_hash
|
| 105 |
+
- runs/ctt_residual_base_context_obs_rollout_train_cal_seed2/metrics.json: missing data_hash
|
| 106 |
+
- runs/ctt_residual_base_context_obs_rollout_train_cal_seed2/metrics.json: missing split_hash
|
| 107 |
+
- runs/ctt_residual_base_context_obs_rollout_val69_seed0/metrics.json: missing data_hash
|
| 108 |
+
- runs/ctt_residual_base_context_obs_rollout_val69_seed0/metrics.json: missing split_hash
|
| 109 |
+
- runs/ctt_residual_base_context_obs_rollout_val69_seed1/metrics.json: missing data_hash
|
| 110 |
+
- runs/ctt_residual_base_context_obs_rollout_val69_seed1/metrics.json: missing split_hash
|
| 111 |
+
- runs/ctt_residual_base_context_obs_rollout_val69_seed2/metrics.json: missing data_hash
|
| 112 |
+
- runs/ctt_residual_base_context_obs_rollout_val69_seed2/metrics.json: missing split_hash
|
| 113 |
+
- runs/ctt_residual_rollout_direct_smoke_seed0_v3/metrics.json: missing data_hash
|
| 114 |
+
- runs/ctt_residual_rollout_direct_smoke_seed0_v3/metrics.json: missing split_hash
|
| 115 |
+
- runs/ctt_residual_rollout_test_seed0/metrics.json: missing data_hash
|
| 116 |
+
- runs/ctt_residual_rollout_test_seed0/metrics.json: missing split_hash
|
| 117 |
+
- runs/ctt_residual_rollout_test_seed1/metrics.json: missing data_hash
|
| 118 |
+
- runs/ctt_residual_rollout_test_seed1/metrics.json: missing split_hash
|
| 119 |
+
- runs/ctt_residual_rollout_test_seed2/metrics.json: missing data_hash
|
| 120 |
+
- runs/ctt_residual_rollout_test_seed2/metrics.json: missing split_hash
|
| 121 |
+
- runs/ctt_residual_rollout_val16_seed0/metrics.json: missing data_hash
|
| 122 |
+
- runs/ctt_residual_rollout_val16_seed0/metrics.json: missing split_hash
|
| 123 |
+
- runs/ctt_residual_rollout_val69_seed0/metrics.json: missing data_hash
|
| 124 |
+
- runs/ctt_residual_rollout_val69_seed0/metrics.json: missing split_hash
|
| 125 |
+
- runs/ctt_residual_rollout_val69_seed1/metrics.json: missing data_hash
|
| 126 |
+
- runs/ctt_residual_rollout_val69_seed1/metrics.json: missing split_hash
|
| 127 |
+
- runs/ctt_residual_rollout_val69_seed2/metrics.json: missing data_hash
|
| 128 |
+
- runs/ctt_residual_rollout_val69_seed2/metrics.json: missing split_hash
|
| 129 |
+
- runs/ctt_test_rollout_comparison/metrics.json: missing data_hash
|
| 130 |
+
- runs/ctt_test_rollout_comparison/metrics.json: missing split_hash
|
| 131 |
+
- runs/ctt_val16_rollout_comparison/metrics.json: missing data_hash
|
| 132 |
+
- runs/ctt_val16_rollout_comparison/metrics.json: missing split_hash
|
| 133 |
+
- runs/ctt_val_rollout_comparison/metrics.json: missing data_hash
|
| 134 |
+
- runs/ctt_val_rollout_comparison/metrics.json: missing split_hash
|
| 135 |
+
- runs/tangent_reconstruction/metrics.json: missing data_hash
|
| 136 |
+
- runs/tangent_reconstruction/metrics.json: missing split_hash
|
workspace/runs/summary_ctt.csv
CHANGED
|
@@ -1,5 +1,6 @@
|
|
| 1 |
method,run_path,clean,status,gate,base_success,success,proposal_oracle,hidden_oracle,support_gap,selector_gap,success_support_gap,success_selector_gap,outcome_ptr,pptc_0p20,pptc_0p40,negative_near_0p20,calibration_ece,seeds,ci
|
| 2 |
V0 residual transport diagnostic,runs/reproduce_v0,yes,measured diagnostic baseline,legacy diagnostic,0.2974,0.3890,0.4435,n/a,0.1264,0.0545,0.1264,0.0545,n/a,0.2366,0.5269,0.0533,n/a,3,see run
|
|
|
|
| 3 |
CTT residual base-context-observation,runs/ctt_base_context_obs_nonlinear_dominance_basic_positive_train_to_test,yes,train feature=base_context_obs,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,120,see run
|
| 4 |
CTT residual base-context-observation,runs/ctt_base_context_obs_nonlinear_dominance_context_success_train_to_test,yes,train feature=base_context_obs,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,120,see run
|
| 5 |
CTT residual base-context-observation,runs/ctt_base_context_obs_nonlinear_dominance_context_tangent_positive_train_to_test,yes,train feature=base_context_obs,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,120,see run
|
|
@@ -15,6 +16,25 @@ CTT gated residual,runs/ctt_gated_residual_full_seed2_val_proxy,yes,proxy,not me
|
|
| 15 |
CTT gated residual,runs/ctt_gated_residual_smoke,yes,train,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,unknown,see run
|
| 16 |
CTT gated residual,runs/ctt_gated_residual_smoke_proxy,yes,proxy,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.5000,0.7500,0.0573,n/a,unknown,see run
|
| 17 |
CTT gated residual,runs/ctt_gated_residual_val_proxy,yes,proxy,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.1739,0.5362,0.0281,n/a,47,see run
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 18 |
CTT residual test measured,runs/ctt_residual_base_context_obs_rollout_test_seed0,yes,measured rollout K=8,"fail:selected<47.45,support_gap>7,selector_gap>3",0.2917,0.2500,0.5417,0.7292,0.4736,0.5645,0.2500,0.2917,0.4792,n/a,n/a,n/a,0.2485,40,see measured_metrics
|
| 19 |
CTT residual test measured,runs/ctt_residual_base_context_obs_rollout_test_seed1,yes,measured rollout K=8,"fail:selected<47.45,support_gap>7,selector_gap>3",0.2917,0.2917,0.5000,0.7292,0.5097,0.3980,0.2708,0.2083,0.5625,n/a,n/a,n/a,0.2274,40,see measured_metrics
|
| 20 |
CTT residual test measured,runs/ctt_residual_base_context_obs_rollout_test_seed2,yes,measured rollout K=8,"fail:selected<47.45,support_gap>7,selector_gap>3",0.2917,0.2708,0.5000,0.7292,0.4822,0.4816,0.2708,0.2292,0.5625,n/a,n/a,n/a,0.1980,40,see measured_metrics
|
|
@@ -57,6 +77,8 @@ local atlas,runs/local_atlas_val_proxy,yes,proxy baseline,not measured,n/a,n/a,n
|
|
| 57 |
ctt residual full,runs/ctt_val_proxy_comparison,yes,proxy aggregate gate=True,proxy=True,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.1981,0.6087,0.0296,n/a,3,aggregate over run means
|
| 58 |
ctt residual base context,runs/ctt_val_proxy_comparison,yes,proxy aggregate gate=True,proxy=True,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.1739,0.6232,0.0182,n/a,1,aggregate over run means
|
| 59 |
ctt residual base context obs,runs/ctt_val_proxy_comparison,yes,proxy aggregate gate=True,proxy=True,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2464,0.6425,0.0343,n/a,3,aggregate over run means
|
|
|
|
|
|
|
| 60 |
ctt gated residual full,runs/ctt_val_proxy_comparison,yes,proxy aggregate gate=False,proxy=False,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2319,0.6135,0.0527,n/a,3,aggregate over run means
|
| 61 |
Utility energy smoke,runs/utility_energy_smoke,train-only,calibration diagnostic,not deployment,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.1316,train split,see run
|
| 62 |
CTT residual test utility-energy dominance,runs/ctt_dominance_utility_energy_smoke_val_to_test,yes,measured dominance K=8 coverage=0.2847 fallback=0.7153,"fail:selected<47.45,support_gap>7,selector_gap>3",0.2847,0.2222,0.5069,0.7292,0.4833,0.5744,0.2639,0.3264,0.5278,n/a,n/a,n/a,n/a,3,see run
|
|
|
|
| 1 |
method,run_path,clean,status,gate,base_success,success,proposal_oracle,hidden_oracle,support_gap,selector_gap,success_support_gap,success_selector_gap,outcome_ptr,pptc_0p20,pptc_0p40,negative_near_0p20,calibration_ece,seeds,ci
|
| 2 |
V0 residual transport diagnostic,runs/reproduce_v0,yes,measured diagnostic baseline,legacy diagnostic,0.2974,0.3890,0.4435,n/a,0.1264,0.0545,0.1264,0.0545,n/a,0.2366,0.5269,0.0533,n/a,3,see run
|
| 3 |
+
CTT residual validation measured aggregate,runs/ctt_base_context_obj_val_rollout_comparison,yes,measured rollout K=8,"fail:selected<47.45,oracle<50,support_gap>7,selector_gap>3",0.2754,0.2029,0.3816,0.6667,0.5369,0.3501,0.2947,0.1787,0.5024,n/a,n/a,n/a,0.2110,3,see measured_metrics
|
| 4 |
CTT residual base-context-observation,runs/ctt_base_context_obs_nonlinear_dominance_basic_positive_train_to_test,yes,train feature=base_context_obs,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,120,see run
|
| 5 |
CTT residual base-context-observation,runs/ctt_base_context_obs_nonlinear_dominance_context_success_train_to_test,yes,train feature=base_context_obs,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,120,see run
|
| 6 |
CTT residual base-context-observation,runs/ctt_base_context_obs_nonlinear_dominance_context_tangent_positive_train_to_test,yes,train feature=base_context_obs,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,120,see run
|
|
|
|
| 16 |
CTT gated residual,runs/ctt_gated_residual_smoke,yes,train,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,unknown,see run
|
| 17 |
CTT gated residual,runs/ctt_gated_residual_smoke_proxy,yes,proxy,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.5000,0.7500,0.0573,n/a,unknown,see run
|
| 18 |
CTT gated residual,runs/ctt_gated_residual_val_proxy,yes,proxy,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.1739,0.5362,0.0281,n/a,47,see run
|
| 19 |
+
CTT residual validation measured,runs/ctt_residual_base_context_obj_rollout_val69_seed0,yes,measured rollout K=8,"fail:selected<47.45,oracle<50,support_gap>7,selector_gap>3",0.2754,0.1739,0.3478,0.6667,0.6088,0.3487,0.3333,0.1739,0.4638,n/a,n/a,n/a,0.2266,47,see measured_metrics
|
| 20 |
+
CTT residual validation measured,runs/ctt_residual_base_context_obj_rollout_val69_seed1,yes,measured rollout K=8,"fail:selected<47.45,oracle<50,support_gap>7,selector_gap>3",0.2754,0.1884,0.3768,0.6667,0.5194,0.3633,0.2899,0.1884,0.5362,n/a,n/a,n/a,0.1740,47,see measured_metrics
|
| 21 |
+
CTT residual validation measured,runs/ctt_residual_base_context_obj_rollout_val69_seed2,yes,measured rollout K=8,"fail:selected<47.45,oracle<50,support_gap>7,selector_gap>3",0.2754,0.2464,0.4203,0.6667,0.4826,0.3382,0.2609,0.1739,0.5072,n/a,n/a,n/a,0.2326,47,see measured_metrics
|
| 22 |
+
CTT residual base-context,runs/ctt_residual_base_context_obj_seed0,yes,train feature=base_context_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,unknown,see run
|
| 23 |
+
CTT residual base-context,runs/ctt_residual_base_context_obj_seed0_val_proxy,yes,proxy feature=base_context_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2609,0.6667,0.0351,n/a,47,see run
|
| 24 |
+
CTT residual base-context,runs/ctt_residual_base_context_obj_seed1,yes,train feature=base_context_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,unknown,see run
|
| 25 |
+
CTT residual base-context,runs/ctt_residual_base_context_obj_seed1_val_proxy,yes,proxy feature=base_context_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2029,0.6522,0.0484,n/a,47,see run
|
| 26 |
+
CTT residual base-context,runs/ctt_residual_base_context_obj_seed2,yes,train feature=base_context_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,unknown,see run
|
| 27 |
+
CTT residual base-context,runs/ctt_residual_base_context_obj_seed2_val_proxy,yes,proxy feature=base_context_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2174,0.6087,0.0306,n/a,47,see run
|
| 28 |
+
CTT residual base-context,runs/ctt_residual_base_context_obj_smoke_seed0,yes,train feature=base_context_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,unknown,see run
|
| 29 |
+
CTT residual base-context,runs/ctt_residual_base_context_obj_smoke_seed0_val_proxy,yes,proxy feature=base_context_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.0000,0.3750,0.0000,n/a,8,see run
|
| 30 |
+
CTT residual base-context-observation,runs/ctt_residual_base_context_obs_obj_seed0,yes,train feature=base_context_obs_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,unknown,see run
|
| 31 |
+
CTT residual base-context-observation,runs/ctt_residual_base_context_obs_obj_seed0_val_proxy,yes,proxy feature=base_context_obs_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.1449,0.6232,0.0031,n/a,47,see run
|
| 32 |
+
CTT residual base-context-observation,runs/ctt_residual_base_context_obs_obj_seed1,yes,train feature=base_context_obs_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,unknown,see run
|
| 33 |
+
CTT residual base-context-observation,runs/ctt_residual_base_context_obs_obj_seed1_val_proxy,yes,proxy feature=base_context_obs_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2319,0.6522,0.0304,n/a,47,see run
|
| 34 |
+
CTT residual base-context-observation,runs/ctt_residual_base_context_obs_obj_seed2,yes,train feature=base_context_obs_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,unknown,see run
|
| 35 |
+
CTT residual base-context-observation,runs/ctt_residual_base_context_obs_obj_seed2_val_proxy,yes,proxy feature=base_context_obs_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2464,0.6522,0.0521,n/a,47,see run
|
| 36 |
+
CTT residual base-context-observation,runs/ctt_residual_base_context_obs_obj_smoke_seed0,yes,train feature=base_context_obs_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,unknown,see run
|
| 37 |
+
CTT residual base-context-observation,runs/ctt_residual_base_context_obs_obj_smoke_seed0_val_proxy,yes,proxy feature=base_context_obs_obj,not measured,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.0000,0.3750,0.0000,n/a,8,see run
|
| 38 |
CTT residual test measured,runs/ctt_residual_base_context_obs_rollout_test_seed0,yes,measured rollout K=8,"fail:selected<47.45,support_gap>7,selector_gap>3",0.2917,0.2500,0.5417,0.7292,0.4736,0.5645,0.2500,0.2917,0.4792,n/a,n/a,n/a,0.2485,40,see measured_metrics
|
| 39 |
CTT residual test measured,runs/ctt_residual_base_context_obs_rollout_test_seed1,yes,measured rollout K=8,"fail:selected<47.45,support_gap>7,selector_gap>3",0.2917,0.2917,0.5000,0.7292,0.5097,0.3980,0.2708,0.2083,0.5625,n/a,n/a,n/a,0.2274,40,see measured_metrics
|
| 40 |
CTT residual test measured,runs/ctt_residual_base_context_obs_rollout_test_seed2,yes,measured rollout K=8,"fail:selected<47.45,support_gap>7,selector_gap>3",0.2917,0.2708,0.5000,0.7292,0.4822,0.4816,0.2708,0.2292,0.5625,n/a,n/a,n/a,0.1980,40,see measured_metrics
|
|
|
|
| 77 |
ctt residual full,runs/ctt_val_proxy_comparison,yes,proxy aggregate gate=True,proxy=True,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.1981,0.6087,0.0296,n/a,3,aggregate over run means
|
| 78 |
ctt residual base context,runs/ctt_val_proxy_comparison,yes,proxy aggregate gate=True,proxy=True,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.1739,0.6232,0.0182,n/a,1,aggregate over run means
|
| 79 |
ctt residual base context obs,runs/ctt_val_proxy_comparison,yes,proxy aggregate gate=True,proxy=True,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2464,0.6425,0.0343,n/a,3,aggregate over run means
|
| 80 |
+
ctt residual base context obj,runs/ctt_val_proxy_comparison,yes,proxy aggregate gate=True,proxy=True,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2271,0.6425,0.0380,n/a,3,aggregate over run means
|
| 81 |
+
ctt residual base context obs obj,runs/ctt_val_proxy_comparison,yes,proxy aggregate gate=True,proxy=True,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2077,0.6425,0.0285,n/a,3,aggregate over run means
|
| 82 |
ctt gated residual full,runs/ctt_val_proxy_comparison,yes,proxy aggregate gate=False,proxy=False,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.2319,0.6135,0.0527,n/a,3,aggregate over run means
|
| 83 |
Utility energy smoke,runs/utility_energy_smoke,train-only,calibration diagnostic,not deployment,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,n/a,0.1316,train split,see run
|
| 84 |
CTT residual test utility-energy dominance,runs/ctt_dominance_utility_energy_smoke_val_to_test,yes,measured dominance K=8 coverage=0.2847 fallback=0.7153,"fail:selected<47.45,support_gap>7,selector_gap>3",0.2847,0.2222,0.5069,0.7292,0.4833,0.5744,0.2639,0.3264,0.5278,n/a,n/a,n/a,n/a,3,see run
|
workspace/runs/summary_ctt.md
CHANGED
|
@@ -3,6 +3,7 @@
|
|
| 3 |
| Method | Status | Base | Selected | Proposal oracle | Hidden oracle | Utility support gap | Utility selector gap | Success support gap | Success selector gap | OutcomePTR | PPTC@0.20 | PPTC@0.40 | NegativeNear@0.20 | Calibration ECE | Run |
|
| 4 |
| --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | --- |
|
| 5 |
| V0 residual transport diagnostic | measured diagnostic baseline; gate: legacy diagnostic | 0.2974 | 0.3890 | 0.4435 | n/a | 0.1264 | 0.0545 | 0.1264 | 0.0545 | n/a | 0.2366 | 0.5269 | 0.0533 | n/a | `runs/reproduce_v0` |
|
|
|
|
| 6 |
| CTT residual base-context-observation | train feature=base_context_obs; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_base_context_obs_nonlinear_dominance_basic_positive_train_to_test` |
|
| 7 |
| CTT residual base-context-observation | train feature=base_context_obs; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_base_context_obs_nonlinear_dominance_context_success_train_to_test` |
|
| 8 |
| CTT residual base-context-observation | train feature=base_context_obs; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_base_context_obs_nonlinear_dominance_context_tangent_positive_train_to_test` |
|
|
@@ -18,6 +19,25 @@
|
|
| 18 |
| CTT gated residual | train; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_gated_residual_smoke` |
|
| 19 |
| CTT gated residual | proxy; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.5000 | 0.7500 | 0.0573 | n/a | `runs/ctt_gated_residual_smoke_proxy` |
|
| 20 |
| CTT gated residual | proxy; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.1739 | 0.5362 | 0.0281 | n/a | `runs/ctt_gated_residual_val_proxy` |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21 |
| CTT residual test measured | measured rollout K=8; gate: fail:selected<47.45,support_gap>7,selector_gap>3 | 0.2917 | 0.2500 | 0.5417 | 0.7292 | 0.4736 | 0.5645 | 0.2500 | 0.2917 | 0.4792 | n/a | n/a | n/a | 0.2485 | `runs/ctt_residual_base_context_obs_rollout_test_seed0` |
|
| 22 |
| CTT residual test measured | measured rollout K=8; gate: fail:selected<47.45,support_gap>7,selector_gap>3 | 0.2917 | 0.2917 | 0.5000 | 0.7292 | 0.5097 | 0.3980 | 0.2708 | 0.2083 | 0.5625 | n/a | n/a | n/a | 0.2274 | `runs/ctt_residual_base_context_obs_rollout_test_seed1` |
|
| 23 |
| CTT residual test measured | measured rollout K=8; gate: fail:selected<47.45,support_gap>7,selector_gap>3 | 0.2917 | 0.2708 | 0.5000 | 0.7292 | 0.4822 | 0.4816 | 0.2708 | 0.2292 | 0.5625 | n/a | n/a | n/a | 0.1980 | `runs/ctt_residual_base_context_obs_rollout_test_seed2` |
|
|
@@ -60,6 +80,8 @@
|
|
| 60 |
| ctt residual full | proxy aggregate gate=True; gate: proxy=True | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.1981 | 0.6087 | 0.0296 | n/a | `runs/ctt_val_proxy_comparison` |
|
| 61 |
| ctt residual base context | proxy aggregate gate=True; gate: proxy=True | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.1739 | 0.6232 | 0.0182 | n/a | `runs/ctt_val_proxy_comparison` |
|
| 62 |
| ctt residual base context obs | proxy aggregate gate=True; gate: proxy=True | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2464 | 0.6425 | 0.0343 | n/a | `runs/ctt_val_proxy_comparison` |
|
|
|
|
|
|
|
| 63 |
| ctt gated residual full | proxy aggregate gate=False; gate: proxy=False | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2319 | 0.6135 | 0.0527 | n/a | `runs/ctt_val_proxy_comparison` |
|
| 64 |
| Utility energy smoke | calibration diagnostic; gate: not deployment | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.1316 | `runs/utility_energy_smoke` |
|
| 65 |
| CTT residual test utility-energy dominance | measured dominance K=8 coverage=0.2847 fallback=0.7153; gate: fail:selected<47.45,support_gap>7,selector_gap>3 | 0.2847 | 0.2222 | 0.5069 | 0.7292 | 0.4833 | 0.5744 | 0.2639 | 0.3264 | 0.5278 | n/a | n/a | n/a | n/a | `runs/ctt_dominance_utility_energy_smoke_val_to_test` |
|
|
|
|
| 3 |
| Method | Status | Base | Selected | Proposal oracle | Hidden oracle | Utility support gap | Utility selector gap | Success support gap | Success selector gap | OutcomePTR | PPTC@0.20 | PPTC@0.40 | NegativeNear@0.20 | Calibration ECE | Run |
|
| 4 |
| --- | --- | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | ---: | --- |
|
| 5 |
| V0 residual transport diagnostic | measured diagnostic baseline; gate: legacy diagnostic | 0.2974 | 0.3890 | 0.4435 | n/a | 0.1264 | 0.0545 | 0.1264 | 0.0545 | n/a | 0.2366 | 0.5269 | 0.0533 | n/a | `runs/reproduce_v0` |
|
| 6 |
+
| CTT residual validation measured aggregate | measured rollout K=8; gate: fail:selected<47.45,oracle<50,support_gap>7,selector_gap>3 | 0.2754 | 0.2029 | 0.3816 | 0.6667 | 0.5369 | 0.3501 | 0.2947 | 0.1787 | 0.5024 | n/a | n/a | n/a | 0.2110 | `runs/ctt_base_context_obj_val_rollout_comparison` |
|
| 7 |
| CTT residual base-context-observation | train feature=base_context_obs; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_base_context_obs_nonlinear_dominance_basic_positive_train_to_test` |
|
| 8 |
| CTT residual base-context-observation | train feature=base_context_obs; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_base_context_obs_nonlinear_dominance_context_success_train_to_test` |
|
| 9 |
| CTT residual base-context-observation | train feature=base_context_obs; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_base_context_obs_nonlinear_dominance_context_tangent_positive_train_to_test` |
|
|
|
|
| 19 |
| CTT gated residual | train; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_gated_residual_smoke` |
|
| 20 |
| CTT gated residual | proxy; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.5000 | 0.7500 | 0.0573 | n/a | `runs/ctt_gated_residual_smoke_proxy` |
|
| 21 |
| CTT gated residual | proxy; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.1739 | 0.5362 | 0.0281 | n/a | `runs/ctt_gated_residual_val_proxy` |
|
| 22 |
+
| CTT residual validation measured | measured rollout K=8; gate: fail:selected<47.45,oracle<50,support_gap>7,selector_gap>3 | 0.2754 | 0.1739 | 0.3478 | 0.6667 | 0.6088 | 0.3487 | 0.3333 | 0.1739 | 0.4638 | n/a | n/a | n/a | 0.2266 | `runs/ctt_residual_base_context_obj_rollout_val69_seed0` |
|
| 23 |
+
| CTT residual validation measured | measured rollout K=8; gate: fail:selected<47.45,oracle<50,support_gap>7,selector_gap>3 | 0.2754 | 0.1884 | 0.3768 | 0.6667 | 0.5194 | 0.3633 | 0.2899 | 0.1884 | 0.5362 | n/a | n/a | n/a | 0.1740 | `runs/ctt_residual_base_context_obj_rollout_val69_seed1` |
|
| 24 |
+
| CTT residual validation measured | measured rollout K=8; gate: fail:selected<47.45,oracle<50,support_gap>7,selector_gap>3 | 0.2754 | 0.2464 | 0.4203 | 0.6667 | 0.4826 | 0.3382 | 0.2609 | 0.1739 | 0.5072 | n/a | n/a | n/a | 0.2326 | `runs/ctt_residual_base_context_obj_rollout_val69_seed2` |
|
| 25 |
+
| CTT residual base-context | train feature=base_context_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_residual_base_context_obj_seed0` |
|
| 26 |
+
| CTT residual base-context | proxy feature=base_context_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2609 | 0.6667 | 0.0351 | n/a | `runs/ctt_residual_base_context_obj_seed0_val_proxy` |
|
| 27 |
+
| CTT residual base-context | train feature=base_context_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_residual_base_context_obj_seed1` |
|
| 28 |
+
| CTT residual base-context | proxy feature=base_context_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2029 | 0.6522 | 0.0484 | n/a | `runs/ctt_residual_base_context_obj_seed1_val_proxy` |
|
| 29 |
+
| CTT residual base-context | train feature=base_context_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_residual_base_context_obj_seed2` |
|
| 30 |
+
| CTT residual base-context | proxy feature=base_context_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2174 | 0.6087 | 0.0306 | n/a | `runs/ctt_residual_base_context_obj_seed2_val_proxy` |
|
| 31 |
+
| CTT residual base-context | train feature=base_context_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_residual_base_context_obj_smoke_seed0` |
|
| 32 |
+
| CTT residual base-context | proxy feature=base_context_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.0000 | 0.3750 | 0.0000 | n/a | `runs/ctt_residual_base_context_obj_smoke_seed0_val_proxy` |
|
| 33 |
+
| CTT residual base-context-observation | train feature=base_context_obs_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_residual_base_context_obs_obj_seed0` |
|
| 34 |
+
| CTT residual base-context-observation | proxy feature=base_context_obs_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.1449 | 0.6232 | 0.0031 | n/a | `runs/ctt_residual_base_context_obs_obj_seed0_val_proxy` |
|
| 35 |
+
| CTT residual base-context-observation | train feature=base_context_obs_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_residual_base_context_obs_obj_seed1` |
|
| 36 |
+
| CTT residual base-context-observation | proxy feature=base_context_obs_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2319 | 0.6522 | 0.0304 | n/a | `runs/ctt_residual_base_context_obs_obj_seed1_val_proxy` |
|
| 37 |
+
| CTT residual base-context-observation | train feature=base_context_obs_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_residual_base_context_obs_obj_seed2` |
|
| 38 |
+
| CTT residual base-context-observation | proxy feature=base_context_obs_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2464 | 0.6522 | 0.0521 | n/a | `runs/ctt_residual_base_context_obs_obj_seed2_val_proxy` |
|
| 39 |
+
| CTT residual base-context-observation | train feature=base_context_obs_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | `runs/ctt_residual_base_context_obs_obj_smoke_seed0` |
|
| 40 |
+
| CTT residual base-context-observation | proxy feature=base_context_obs_obj; gate: not measured | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.0000 | 0.3750 | 0.0000 | n/a | `runs/ctt_residual_base_context_obs_obj_smoke_seed0_val_proxy` |
|
| 41 |
| CTT residual test measured | measured rollout K=8; gate: fail:selected<47.45,support_gap>7,selector_gap>3 | 0.2917 | 0.2500 | 0.5417 | 0.7292 | 0.4736 | 0.5645 | 0.2500 | 0.2917 | 0.4792 | n/a | n/a | n/a | 0.2485 | `runs/ctt_residual_base_context_obs_rollout_test_seed0` |
|
| 42 |
| CTT residual test measured | measured rollout K=8; gate: fail:selected<47.45,support_gap>7,selector_gap>3 | 0.2917 | 0.2917 | 0.5000 | 0.7292 | 0.5097 | 0.3980 | 0.2708 | 0.2083 | 0.5625 | n/a | n/a | n/a | 0.2274 | `runs/ctt_residual_base_context_obs_rollout_test_seed1` |
|
| 43 |
| CTT residual test measured | measured rollout K=8; gate: fail:selected<47.45,support_gap>7,selector_gap>3 | 0.2917 | 0.2708 | 0.5000 | 0.7292 | 0.4822 | 0.4816 | 0.2708 | 0.2292 | 0.5625 | n/a | n/a | n/a | 0.1980 | `runs/ctt_residual_base_context_obs_rollout_test_seed2` |
|
|
|
|
| 80 |
| ctt residual full | proxy aggregate gate=True; gate: proxy=True | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.1981 | 0.6087 | 0.0296 | n/a | `runs/ctt_val_proxy_comparison` |
|
| 81 |
| ctt residual base context | proxy aggregate gate=True; gate: proxy=True | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.1739 | 0.6232 | 0.0182 | n/a | `runs/ctt_val_proxy_comparison` |
|
| 82 |
| ctt residual base context obs | proxy aggregate gate=True; gate: proxy=True | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2464 | 0.6425 | 0.0343 | n/a | `runs/ctt_val_proxy_comparison` |
|
| 83 |
+
| ctt residual base context obj | proxy aggregate gate=True; gate: proxy=True | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2271 | 0.6425 | 0.0380 | n/a | `runs/ctt_val_proxy_comparison` |
|
| 84 |
+
| ctt residual base context obs obj | proxy aggregate gate=True; gate: proxy=True | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2077 | 0.6425 | 0.0285 | n/a | `runs/ctt_val_proxy_comparison` |
|
| 85 |
| ctt gated residual full | proxy aggregate gate=False; gate: proxy=False | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.2319 | 0.6135 | 0.0527 | n/a | `runs/ctt_val_proxy_comparison` |
|
| 86 |
| Utility energy smoke | calibration diagnostic; gate: not deployment | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | 0.1316 | `runs/utility_energy_smoke` |
|
| 87 |
| CTT residual test utility-energy dominance | measured dominance K=8 coverage=0.2847 fallback=0.7153; gate: fail:selected<47.45,support_gap>7,selector_gap>3 | 0.2847 | 0.2222 | 0.5069 | 0.7292 | 0.4833 | 0.5744 | 0.2639 | 0.3264 | 0.5278 | n/a | n/a | n/a | n/a | `runs/ctt_dominance_utility_energy_smoke_val_to_test` |
|