auto-sync 2026-07-04T08:12:11Z workspace (part 2)

workspace/runs/ctt_base_context_obs_learned_dominance_bundle_envclip_k16_train_to_test/eval.log

@@ -0,0 +1,3 @@
+evaluated learned fallback rule on held-out measured rollout rows
+eval_input=runs/ctt_base_context_obs_test_envclip_k16_rollout_comparison/combined_measured_candidates.json
+num_eval_rows=144

workspace/runs/ctt_base_context_obs_learned_dominance_bundle_envclip_k16_train_to_test/metrics_by_task.json

@@ -0,0 +1,137 @@
+{
+  "LiftPegUpright-v1": {
+    "base_success": 0.2,
+    "base_utility": 0.9498705387115478,
+    "coverage": 0.6666666666666666,
+    "execute_generated": 0.6666666666666666,
+    "fallback_rate": 0.3333333333333333,
+    "hidden_chart_oracle_success": 1.0,
+    "hidden_chart_oracle_utility": 2.0,
+    "outcome_ptr": 0.5333333333333333,
+    "pairwise_causal_calibration_accuracy": 0.4784086828204475,
+    "pairwise_causal_calibration_confidence": 0.5335629666344366,
+    "pairwise_causal_calibration_ece": 0.1274266477632942,
+    "pairwise_causal_calibration_pairs": 94.8,
+    "predicted_margin": 0.03252291650109252,
+    "proposal_oracle_success": 0.6,
+    "proposal_oracle_utility": 1.4122399787108104,
+    "selected_candidate_index": 7.333333333333333,
+    "selected_success": 0.13333333333333333,
+    "selected_success_gain_over_base": -0.06666666666666667,
+    "selected_utility": 0.7594087203343709,
+    "selected_utility_gain_over_base": -0.19046181837717693,
+    "selector_regret": 0.6795656750599544,
+    "success_selector_gap": 0.4666666666666667,
+    "success_support_gap": 0.4,
+    "support_gap": 0.5877600212891897,
+    "tau": -0.07385947182024077
+  },
+  "PickCube-v1": {
+    "base_success": 0.0,
+    "base_utility": 0.5839591026306152,
+    "coverage": 0.0,
+    "execute_generated": 0.0,
+    "fallback_rate": 1.0,
+    "hidden_chart_oracle_success": 1.0,
+    "hidden_chart_oracle_utility": 2.0,
+    "outcome_ptr": 0.0,
+    "pairwise_causal_calibration_accuracy": 0.5370370370370371,
+    "pairwise_causal_calibration_confidence": 0.5273646918080489,
+    "pairwise_causal_calibration_ece": 0.11910009801460207,
+    "pairwise_causal_calibration_pairs": 36.0,
+    "predicted_margin": -0.25931321700031695,
+    "proposal_oracle_success": 0.0,
+    "proposal_oracle_utility": 0.13425676027933756,
+    "selected_candidate_index": 3.6666666666666665,
+    "selected_success": 0.0,
+    "selected_success_gain_over_base": 0.0,
+    "selected_utility": 0.5839591026306152,
+    "selected_utility_gain_over_base": 0.0,
+    "selector_regret": 0.0,
+    "success_selector_gap": 0.0,
+    "success_support_gap": 1.0,
+    "support_gap": 1.8657432397206624,
+    "tau": 0.02896384295113541
+  },
+  "PullCube-v1": {
+    "base_success": 0.36,
+    "base_utility": 0.7751704612234608,
+    "coverage": 0.7333333333333333,
+    "execute_generated": 0.7333333333333333,
+    "fallback_rate": 0.26666666666666666,
+    "hidden_chart_oracle_success": 0.52,
+    "hidden_chart_oracle_utility": 1.1467878860235214,
+    "outcome_ptr": 0.4666666666666667,
+    "pairwise_causal_calibration_accuracy": 0.49303016307826086,
+    "pairwise_causal_calibration_confidence": 0.5282915670267949,
+    "pairwise_causal_calibration_ece": 0.09823916112294553,
+    "pairwise_causal_calibration_pairs": 92.32,
+    "predicted_margin": 0.0830997467005694,
+    "proposal_oracle_success": 0.5333333333333333,
+    "proposal_oracle_utility": 1.1093241602430741,
+    "selected_candidate_index": 6.16,
+    "selected_success": 0.29333333333333333,
+    "selected_success_gain_over_base": -0.06666666666666667,
+    "selected_utility": 0.6347246278480937,
+    "selected_utility_gain_over_base": -0.14044583337536704,
+    "selector_regret": 0.5004223387533178,
+    "success_selector_gap": 0.25333333333333335,
+    "success_support_gap": 0.02666666666666667,
+    "support_gap": 0.1075624164690574,
+    "tau": -0.07186117033841724
+  },
+  "PushCube-v1": {
+    "base_success": 0.16666666666666666,
+    "base_utility": 0.3924041787783305,
+    "coverage": 0.8333333333333334,
+    "execute_generated": 0.8333333333333334,
+    "fallback_rate": 0.16666666666666666,
+    "hidden_chart_oracle_success": 0.8333333333333334,
+    "hidden_chart_oracle_utility": 1.7015800600250561,
+    "outcome_ptr": 0.8333333333333334,
+    "pairwise_causal_calibration_accuracy": 0.5754796516735593,
+    "pairwise_causal_calibration_confidence": 0.528992677243157,
+    "pairwise_causal_calibration_ece": 0.08128349345238421,
+    "pairwise_causal_calibration_pairs": 97.44444444444444,
+    "predicted_margin": 0.44319861689531703,
+    "proposal_oracle_success": 0.8888888888888888,
+    "proposal_oracle_utility": 1.7857814149724112,
+    "selected_candidate_index": 6.944444444444445,
+    "selected_success": 0.4444444444444444,
+    "selected_success_gain_over_base": 0.2777777777777778,
+    "selected_utility": 0.9508692075808843,
+    "selected_utility_gain_over_base": 0.5584650288025538,
+    "selector_regret": 0.8349122073915269,
+    "success_selector_gap": 0.4444444444444444,
+    "success_support_gap": 0.1111111111111111,
+    "support_gap": 0.21421858502758873,
+    "tau": 0.2874440124493883
+  },
+  "StackCube-v1": {
+    "base_success": 0.2727272727272727,
+    "base_utility": 0.7763947045261209,
+    "coverage": 0.9393939393939394,
+    "execute_generated": 0.9393939393939394,
+    "fallback_rate": 0.06060606060606061,
+    "hidden_chart_oracle_success": 1.0,
+    "hidden_chart_oracle_utility": 2.0,
+    "outcome_ptr": 0.6363636363636364,
+    "pairwise_causal_calibration_accuracy": 0.5307015979194418,
+    "pairwise_causal_calibration_confidence": 0.5283985674194418,
+    "pairwise_causal_calibration_ece": 0.12082192852053648,
+    "pairwise_causal_calibration_pairs": 67.45454545454545,
+    "predicted_margin": -0.05568968191367572,
+    "proposal_oracle_success": 0.5151515151515151,
+    "proposal_oracle_utility": 1.1567951707225856,
+    "selected_candidate_index": 6.424242424242424,
+    "selected_success": 0.21212121212121213,
+    "selected_success_gain_over_base": -0.06060606060606061,
+    "selected_utility": 0.6115993509689966,
+    "selected_utility_gain_over_base": -0.16479535355712427,
+    "selector_regret": 0.5451958197535891,
+    "success_selector_gap": 0.30303030303030304,
+    "success_support_gap": 0.48484848484848486,
+    "support_gap": 0.8432048292774142,
+    "tau": -0.18710182505305803
+  }
+}

workspace/runs/ctt_base_context_obs_learned_dominance_bundle_envclip_k16_train_to_test/table.tex

@@ -0,0 +1,8 @@
+% Auto-generated by scripts/eval_learned_dominance_selector.py
+\begin{tabular}{lrrrrrrrrr}
+\toprule
+Rows & Coverage & Fallback & Base succ. & Selected succ. & Oracle succ. & OutcomePTR & Succ. support gap & Succ. selector gap & Cal. ECE \\
+\midrule
+144 & 0.7708 & 0.2292 & 0.2917 & 0.2708 & 0.5694 & 0.5486 & 0.2014 & 0.3056 & 0.0151 \\
+\bottomrule
+\end{tabular}

workspace/runs/ctt_base_context_obs_learned_dominance_bundle_envclip_k16_train_to_test/train.log

@@ -0,0 +1,5 @@
+trained ridge dominance calibrator on calibration measured rows only
+fit_objective=pointwise
+selected_lambda=0.0
+tau={"LiftPegUpright-v1": -0.073859, "PickCube-v1": 0.028964, "PullCube-v1": -0.071861, "PushCube-v1": 0.287444, "StackCube-v1": -0.187102, "__global__": -0.071861}
+calibration_input=runs/ctt_base_context_obs_train_cal_envclip_k16_rollout_comparison/combined_measured_candidates.json

workspace/runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test/command.txt

@@ -0,0 +1 @@
+python scripts/eval_learned_dominance_selector.py --calibration-input runs/ctt_base_context_obs_train_cal_envclip_k16_rollout_comparison/combined_measured_candidates.json --calibration-target-index data/cil_charts_rgb_refs/train/index.json --eval-input runs/ctt_base_context_obs_test_envclip_k16_rollout_comparison/combined_measured_candidates.json --eval-target-index data/cil_charts_rgb_refs/test/index.json --checkpoint-template runs/ctt_residual_base_context_obs_seed{seed}/model.pt --out-dir runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test --feature-set chart_bundle_consensus --selector-chart-feature-mode base_context_obs --target utility_margin --threshold-scope task --fit-objective pointwise --k 16 --bootstrap-samples 1000 --no-markdown-report

workspace/runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test/config.yaml

@@ -0,0 +1,18 @@
+bootstrap_samples: 1000
+calibration_input: runs/ctt_base_context_obs_train_cal_envclip_k16_rollout_comparison/combined_measured_candidates.json
+calibration_target_index: data/cil_charts_rgb_refs/train/index.json
+checkpoint_template: runs/ctt_residual_base_context_obs_seed{seed}/model.pt
+eval_input: runs/ctt_base_context_obs_test_envclip_k16_rollout_comparison/combined_measured_candidates.json
+eval_target_index: data/cil_charts_rgb_refs/test/index.json
+feature_set: chart_bundle_consensus
+fit_objective: pointwise
+k: 16
+no_markdown_report: True
+out_dir: runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test
+pairwise_weight: 1.0
+ridge_lambdas: 0,0.01,0.1,1,10,100
+selector_chart_feature_mode: base_context_obs
+source_index: None
+success_bonus: 1.0
+target: utility_margin
+threshold_scope: task

workspace/runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test/data_hash.txt

@@ -0,0 +1,7 @@
+{
+  "calibration_input": "d79e275b896eb98caa128470d4614b76960cab12407faa157c583224be7abe80",
+  "calibration_target_index": "67973c50072aa87a3b97533b6292b08a8b4d88a95571dc22e91b6e8fd19e0d85",
+  "eval_input": "541d5a928da5f7ee04ca3dea570e04208830dffde431edfe5b5e38b692634045",
+  "eval_target_index": "dee1b5225485df6624747dd4c5074bcf514685b54017d05e698d2051638fbf36",
+  "source_index": "67973c50072aa87a3b97533b6292b08a8b4d88a95571dc22e91b6e8fd19e0d85"
+}

workspace/runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test/eval.log

@@ -0,0 +1,3 @@
+evaluated learned fallback rule on held-out measured rollout rows
+eval_input=runs/ctt_base_context_obs_test_envclip_k16_rollout_comparison/combined_measured_candidates.json
+num_eval_rows=144

workspace/runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test/git_hash.txt

@@ -0,0 +1 @@
+14e81292e2004925f81e0bb4810f6f144bfd87b7

workspace/runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test/metrics_by_task.json

@@ -0,0 +1,137 @@
+{
+  "LiftPegUpright-v1": {
+    "base_success": 0.2,
+    "base_utility": 0.9498705387115478,
+    "coverage": 0.6,
+    "execute_generated": 0.6,
+    "fallback_rate": 0.4,
+    "hidden_chart_oracle_success": 1.0,
+    "hidden_chart_oracle_utility": 2.0,
+    "outcome_ptr": 0.5333333333333333,
+    "pairwise_causal_calibration_accuracy": 0.54512467498401,
+    "pairwise_causal_calibration_confidence": 0.54564531858913,
+    "pairwise_causal_calibration_ece": 0.11030342503862721,
+    "pairwise_causal_calibration_pairs": 94.8,
+    "predicted_margin": 0.14233301795724493,
+    "proposal_oracle_success": 0.6,
+    "proposal_oracle_utility": 1.4122399787108104,
+    "selected_candidate_index": 7.6,
+    "selected_success": 0.13333333333333333,
+    "selected_success_gain_over_base": -0.06666666666666667,
+    "selected_utility": 0.8185293744007747,
+    "selected_utility_gain_over_base": -0.13134116431077322,
+    "selector_regret": 0.6451034953196844,
+    "success_selector_gap": 0.4666666666666667,
+    "success_support_gap": 0.4,
+    "support_gap": 0.5877600212891897,
+    "tau": 0.0940318405174132
+  },
+  "PickCube-v1": {
+    "base_success": 0.0,
+    "base_utility": 0.5839591026306152,
+    "coverage": 0.0,
+    "execute_generated": 0.0,
+    "fallback_rate": 1.0,
+    "hidden_chart_oracle_success": 1.0,
+    "hidden_chart_oracle_utility": 2.0,
+    "outcome_ptr": 0.0,
+    "pairwise_causal_calibration_accuracy": 0.39814814814814814,
+    "pairwise_causal_calibration_confidence": 0.5607943587261041,
+    "pairwise_causal_calibration_ece": 0.16599791089703783,
+    "pairwise_causal_calibration_pairs": 36.0,
+    "predicted_margin": -0.39039029063840003,
+    "proposal_oracle_success": 0.0,
+    "proposal_oracle_utility": 0.13425676027933756,
+    "selected_candidate_index": 5.0,
+    "selected_success": 0.0,
+    "selected_success_gain_over_base": 0.0,
+    "selected_utility": 0.5839591026306152,
+    "selected_utility_gain_over_base": 0.0,
+    "selector_regret": 0.0,
+    "success_selector_gap": 0.0,
+    "success_support_gap": 1.0,
+    "support_gap": 1.8657432397206624,
+    "tau": 0.3611323919348086
+  },
+  "PullCube-v1": {
+    "base_success": 0.36,
+    "base_utility": 0.7751704612234608,
+    "coverage": 0.6266666666666667,
+    "execute_generated": 0.6266666666666667,
+    "fallback_rate": 0.37333333333333335,
+    "hidden_chart_oracle_success": 0.52,
+    "hidden_chart_oracle_utility": 1.1467878860235214,
+    "outcome_ptr": 0.4666666666666667,
+    "pairwise_causal_calibration_accuracy": 0.4995022986920425,
+    "pairwise_causal_calibration_confidence": 0.5422481003539483,
+    "pairwise_causal_calibration_ece": 0.11105368029396932,
+    "pairwise_causal_calibration_pairs": 92.32,
+    "predicted_margin": 0.18136174528296545,
+    "proposal_oracle_success": 0.5333333333333333,
+    "proposal_oracle_utility": 1.1093241602430741,
+    "selected_candidate_index": 7.1866666666666665,
+    "selected_success": 0.32,
+    "selected_success_gain_over_base": -0.04,
+    "selected_utility": 0.6814040022250265,
+    "selected_utility_gain_over_base": -0.09376645899843425,
+    "selector_regret": 0.5039629975054413,
+    "success_selector_gap": 0.25333333333333335,
+    "success_support_gap": 0.02666666666666667,
+    "support_gap": 0.1075624164690574,
+    "tau": 0.10332537601561143
+  },
+  "PushCube-v1": {
+    "base_success": 0.16666666666666666,
+    "base_utility": 0.3924041787783305,
+    "coverage": 0.6666666666666666,
+    "execute_generated": 0.6666666666666666,
+    "fallback_rate": 0.3333333333333333,
+    "hidden_chart_oracle_success": 0.8333333333333334,
+    "hidden_chart_oracle_utility": 1.7015800600250561,
+    "outcome_ptr": 0.8333333333333334,
+    "pairwise_causal_calibration_accuracy": 0.588120956369862,
+    "pairwise_causal_calibration_confidence": 0.5338991876592735,
+    "pairwise_causal_calibration_ece": 0.08423604919725729,
+    "pairwise_causal_calibration_pairs": 97.44444444444444,
+    "predicted_margin": 0.4582899698155127,
+    "proposal_oracle_success": 0.8888888888888888,
+    "proposal_oracle_utility": 1.7857814149724112,
+    "selected_candidate_index": 7.055555555555555,
+    "selected_success": 0.5,
+    "selected_success_gain_over_base": 0.3333333333333333,
+    "selected_utility": 1.062640056014061,
+    "selected_utility_gain_over_base": 0.6702358772357305,
+    "selector_regret": 0.7231413589583503,
+    "success_selector_gap": 0.3888888888888889,
+    "success_support_gap": 0.1111111111111111,
+    "support_gap": 0.21421858502758873,
+    "tau": 0.35241155802994506
+  },
+  "StackCube-v1": {
+    "base_success": 0.2727272727272727,
+    "base_utility": 0.7763947045261209,
+    "coverage": 0.6060606060606061,
+    "execute_generated": 0.6060606060606061,
+    "fallback_rate": 0.3939393939393939,
+    "hidden_chart_oracle_success": 1.0,
+    "hidden_chart_oracle_utility": 2.0,
+    "outcome_ptr": 0.6363636363636364,
+    "pairwise_causal_calibration_accuracy": 0.5309413874218266,
+    "pairwise_causal_calibration_confidence": 0.5440067619410754,
+    "pairwise_causal_calibration_ece": 0.09869174841720094,
+    "pairwise_causal_calibration_pairs": 67.45454545454545,
+    "predicted_margin": 0.06420154895093595,
+    "proposal_oracle_success": 0.5151515151515151,
+    "proposal_oracle_utility": 1.1567951707225856,
+    "selected_candidate_index": 6.212121212121212,
+    "selected_success": 0.30303030303030304,
+    "selected_success_gain_over_base": 0.030303030303030304,
+    "selected_utility": 0.8023195727304979,
+    "selected_utility_gain_over_base": 0.025924868204376915,
+    "selector_regret": 0.3561627133326097,
+    "success_selector_gap": 0.21212121212121213,
+    "success_support_gap": 0.48484848484848486,
+    "support_gap": 0.8432048292774142,
+    "tau": 0.0009112489076126273
+  }
+}

workspace/runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test/split_hash.txt

@@ -0,0 +1,14 @@
+{
+  "calibration_target_index": {
+    "content_hash": "67f7b4a692f7c7e71da24378dc71e3399c5d35afc14c494869eba6087b765c42",
+    "retrieval_index_allowed": true,
+    "split": "train",
+    "split_hash": "867a94a1c67686c082765671a7bf8d74a9d633085e1b2619c83197c60bf58372"
+  },
+  "eval_target_index": {
+    "content_hash": "0c38d73dc9bc4ab756e12006671b5d12f1a23358b281e54b3f6a9ae3cbf6e9af",
+    "retrieval_index_allowed": false,
+    "split": "test",
+    "split_hash": "84151df9e88c219b76751677558cc5bbbc19082bf358cea7330158e9e403e25f"
+  }
+}

workspace/runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test/table.tex

@@ -0,0 +1,8 @@
+% Auto-generated by scripts/eval_learned_dominance_selector.py
+\begin{tabular}{lrrrrrrrrr}
+\toprule
+Rows & Coverage & Fallback & Base succ. & Selected succ. & Oracle succ. & OutcomePTR & Succ. support gap & Succ. selector gap & Cal. ECE \\
+\midrule
+144 & 0.6111 & 0.3889 & 0.2917 & 0.3125 & 0.5694 & 0.5486 & 0.2014 & 0.2778 & 0.0255 \\
+\bottomrule
+\end{tabular}

workspace/runs/ctt_base_context_obs_learned_dominance_chart_bundle_envclip_k16_train_to_test/train.log

@@ -0,0 +1,5 @@
+trained ridge dominance calibrator on calibration measured rows only
+fit_objective=pointwise
+selected_lambda=10.0
+tau={"LiftPegUpright-v1": 0.094032, "PickCube-v1": 0.361132, "PullCube-v1": 0.103325, "PushCube-v1": 0.352412, "StackCube-v1": 0.000911, "__global__": 0.113201}
+calibration_input=runs/ctt_base_context_obs_train_cal_envclip_k16_rollout_comparison/combined_measured_candidates.json

workspace/runs/ctt_base_context_obs_learned_dominance_chart_source_bundle_envclip_k16_train_to_test/command.txt

@@ -0,0 +1 @@
+python scripts/eval_learned_dominance_selector.py --calibration-input runs/ctt_base_context_obs_train_cal_envclip_k16_rollout_comparison/combined_measured_candidates.json --calibration-target-index data/cil_charts_rgb_refs/train/index.json --eval-input runs/ctt_base_context_obs_test_envclip_k16_rollout_comparison/combined_measured_candidates.json --eval-target-index data/cil_charts_rgb_refs/test/index.json --source-index data/cil_charts_rgb_refs/train/index.json --checkpoint-template runs/ctt_residual_base_context_obs_seed{seed}/model.pt --out-dir runs/ctt_base_context_obs_learned_dominance_chart_source_bundle_envclip_k16_train_to_test --feature-set chart_source_bundle_consensus --selector-chart-feature-mode base_context_obs --target success_weighted_margin --success-bonus 2.0 --threshold-scope task --fit-objective pointwise --k 16 --bootstrap-samples 1000 --no-markdown-report

workspace/runs/ctt_base_context_obs_learned_dominance_chart_source_bundle_envclip_k16_train_to_test/config.yaml

@@ -0,0 +1,18 @@
+bootstrap_samples: 1000
+calibration_input: runs/ctt_base_context_obs_train_cal_envclip_k16_rollout_comparison/combined_measured_candidates.json
+calibration_target_index: data/cil_charts_rgb_refs/train/index.json
+checkpoint_template: runs/ctt_residual_base_context_obs_seed{seed}/model.pt
+eval_input: runs/ctt_base_context_obs_test_envclip_k16_rollout_comparison/combined_measured_candidates.json
+eval_target_index: data/cil_charts_rgb_refs/test/index.json
+feature_set: chart_source_bundle_consensus
+fit_objective: pointwise
+k: 16
+no_markdown_report: True
+out_dir: runs/ctt_base_context_obs_learned_dominance_chart_source_bundle_envclip_k16_train_to_test
+pairwise_weight: 1.0
+ridge_lambdas: 0,0.01,0.1,1,10,100
+selector_chart_feature_mode: base_context_obs
+source_index: data/cil_charts_rgb_refs/train/index.json
+success_bonus: 2.0
+target: success_weighted_margin
+threshold_scope: task

workspace/runs/ctt_base_context_obs_learned_dominance_chart_source_bundle_envclip_k16_train_to_test/data_hash.txt

@@ -0,0 +1,7 @@
+{
+  "calibration_input": "d79e275b896eb98caa128470d4614b76960cab12407faa157c583224be7abe80",
+  "calibration_target_index": "67973c50072aa87a3b97533b6292b08a8b4d88a95571dc22e91b6e8fd19e0d85",
+  "eval_input": "541d5a928da5f7ee04ca3dea570e04208830dffde431edfe5b5e38b692634045",
+  "eval_target_index": "dee1b5225485df6624747dd4c5074bcf514685b54017d05e698d2051638fbf36",
+  "source_index": "67973c50072aa87a3b97533b6292b08a8b4d88a95571dc22e91b6e8fd19e0d85"
+}

workspace/runs/ctt_base_context_obs_learned_dominance_chart_source_bundle_envclip_k16_train_to_test/eval.log

@@ -0,0 +1,3 @@
+evaluated learned fallback rule on held-out measured rollout rows
+eval_input=runs/ctt_base_context_obs_test_envclip_k16_rollout_comparison/combined_measured_candidates.json
+num_eval_rows=144

workspace/runs/ctt_base_context_obs_learned_dominance_chart_source_bundle_envclip_k16_train_to_test/git_hash.txt

@@ -0,0 +1 @@
+14e81292e2004925f81e0bb4810f6f144bfd87b7

workspace/runs/ctt_base_context_obs_learned_dominance_chart_source_bundle_envclip_k16_train_to_test/metrics_by_task.json

@@ -0,0 +1,137 @@
+{
+  "LiftPegUpright-v1": {
+    "base_success": 0.2,
+    "base_utility": 0.9498705387115478,
+    "coverage": 0.6,
+    "execute_generated": 0.6,
+    "fallback_rate": 0.4,
+    "hidden_chart_oracle_success": 1.0,
+    "hidden_chart_oracle_utility": 2.0,
+    "outcome_ptr": 0.5333333333333333,
+    "pairwise_causal_calibration_accuracy": 0.5555136707566375,
+    "pairwise_causal_calibration_confidence": 0.5720692082177231,
+    "pairwise_causal_calibration_ece": 0.11666816996110206,
+    "pairwise_causal_calibration_pairs": 94.8,
+    "predicted_margin": 0.7829724526600295,
+    "proposal_oracle_success": 0.6,
+    "proposal_oracle_utility": 1.4122399787108104,
+    "selected_candidate_index": 5.133333333333334,
+    "selected_success": 0.06666666666666667,
+    "selected_success_gain_over_base": -0.13333333333333333,
+    "selected_utility": 0.7418742964665095,
+    "selected_utility_gain_over_base": -0.20799624224503835,
+    "selector_regret": 0.7217585732539494,
+    "success_selector_gap": 0.5333333333333333,
+    "success_support_gap": 0.4,
+    "support_gap": 0.5877600212891897,
+    "tau": 0.7478358458742457
+  },
+  "PickCube-v1": {
+    "base_success": 0.0,
+    "base_utility": 0.5839591026306152,
+    "coverage": 0.0,
+    "execute_generated": 0.0,
+    "fallback_rate": 1.0,
+    "hidden_chart_oracle_success": 1.0,
+    "hidden_chart_oracle_utility": 2.0,
+    "outcome_ptr": 0.0,
+    "pairwise_causal_calibration_accuracy": 0.3333333333333333,
+    "pairwise_causal_calibration_confidence": 0.6023940721845762,
+    "pairwise_causal_calibration_ece": 0.2745681321943875,
+    "pairwise_causal_calibration_pairs": 36.0,
+    "predicted_margin": -0.12247428613675211,
+    "proposal_oracle_success": 0.0,
+    "proposal_oracle_utility": 0.13425676027933756,
+    "selected_candidate_index": 2.0,
+    "selected_success": 0.0,
+    "selected_success_gain_over_base": 0.0,
+    "selected_utility": 0.5839591026306152,
+    "selected_utility_gain_over_base": 0.0,
+    "selector_regret": 0.0,
+    "success_selector_gap": 0.0,
+    "success_support_gap": 1.0,
+    "support_gap": 1.8657432397206624,
+    "tau": 1.280136139242778
+  },
+  "PullCube-v1": {
+    "base_success": 0.36,
+    "base_utility": 0.7751704612234608,
+    "coverage": 0.52,
+    "execute_generated": 0.52,
+    "fallback_rate": 0.48,
+    "hidden_chart_oracle_success": 0.52,
+    "hidden_chart_oracle_utility": 1.1467878860235214,
+    "outcome_ptr": 0.4666666666666667,
+    "pairwise_causal_calibration_accuracy": 0.5062744920994355,
+    "pairwise_causal_calibration_confidence": 0.5796546075128097,
+    "pairwise_causal_calibration_ece": 0.12491436687229007,
+    "pairwise_causal_calibration_pairs": 92.32,
+    "predicted_margin": 0.9547622553200451,
+    "proposal_oracle_success": 0.5333333333333333,
+    "proposal_oracle_utility": 1.1093241602430741,
+    "selected_candidate_index": 7.266666666666667,
+    "selected_success": 0.32,
+    "selected_success_gain_over_base": -0.04,
+    "selected_utility": 0.6846797956805676,
+    "selected_utility_gain_over_base": -0.09049066554289312,
+    "selector_regret": 0.45504458223469557,
+    "success_selector_gap": 0.22666666666666666,
+    "success_support_gap": 0.02666666666666667,
+    "support_gap": 0.1075624164690574,
+    "tau": 0.9470671281950384
+  },
+  "PushCube-v1": {
+    "base_success": 0.16666666666666666,
+    "base_utility": 0.3924041787783305,
+    "coverage": 0.8888888888888888,
+    "execute_generated": 0.8888888888888888,
+    "fallback_rate": 0.1111111111111111,
+    "hidden_chart_oracle_success": 0.8333333333333334,
+    "hidden_chart_oracle_utility": 1.7015800600250561,
+    "outcome_ptr": 0.8333333333333334,
+    "pairwise_causal_calibration_accuracy": 0.5428451728329846,
+    "pairwise_causal_calibration_confidence": 0.6000598858100837,
+    "pairwise_causal_calibration_ece": 0.15543541117172033,
+    "pairwise_causal_calibration_pairs": 97.44444444444444,
+    "predicted_margin": 1.843134202495152,
+    "proposal_oracle_success": 0.8888888888888888,
+    "proposal_oracle_utility": 1.7857814149724112,
+    "selected_candidate_index": 6.222222222222222,
+    "selected_success": 0.3888888888888889,
+    "selected_success_gain_over_base": 0.2222222222222222,
+    "selected_utility": 0.8422558406988779,
+    "selected_utility_gain_over_base": 0.4498516619205475,
+    "selector_regret": 0.9435255742735333,
+    "success_selector_gap": 0.5,
+    "success_support_gap": 0.1111111111111111,
+    "support_gap": 0.21421858502758873,
+    "tau": 1.2264360294338525
+  },
+  "StackCube-v1": {
+    "base_success": 0.2727272727272727,
+    "base_utility": 0.7763947045261209,
+    "coverage": 0.5151515151515151,
+    "execute_generated": 0.5151515151515151,
+    "fallback_rate": 0.48484848484848486,
+    "hidden_chart_oracle_success": 1.0,
+    "hidden_chart_oracle_utility": 2.0,
+    "outcome_ptr": 0.6363636363636364,
+    "pairwise_causal_calibration_accuracy": 0.5286305231046764,
+    "pairwise_causal_calibration_confidence": 0.5839574678118631,
+    "pairwise_causal_calibration_ece": 0.13914695546983782,
+    "pairwise_causal_calibration_pairs": 67.45454545454545,
+    "predicted_margin": 0.832831201132664,
+    "proposal_oracle_success": 0.5151515151515151,
+    "proposal_oracle_utility": 1.1567951707225856,
+    "selected_candidate_index": 6.606060606060606,
+    "selected_success": 0.30303030303030304,
+    "selected_success_gain_over_base": 0.030303030303030304,
+    "selected_utility": 0.7861830118027601,
+    "selected_utility_gain_over_base": 0.00978830727663907,
+    "selector_regret": 0.37342702303871966,
+    "success_selector_gap": 0.21212121212121213,
+    "success_support_gap": 0.48484848484848486,
+    "support_gap": 0.8432048292774142,
+    "tau": 0.8938079363311323
+  }
+}

workspace/runs/ctt_base_context_obs_learned_dominance_chart_source_bundle_envclip_k16_train_to_test/split_hash.txt

@@ -0,0 +1,14 @@
+{
+  "calibration_target_index": {
+    "content_hash": "67f7b4a692f7c7e71da24378dc71e3399c5d35afc14c494869eba6087b765c42",
+    "retrieval_index_allowed": true,
+    "split": "train",
+    "split_hash": "867a94a1c67686c082765671a7bf8d74a9d633085e1b2619c83197c60bf58372"
+  },
+  "eval_target_index": {
+    "content_hash": "0c38d73dc9bc4ab756e12006671b5d12f1a23358b281e54b3f6a9ae3cbf6e9af",
+    "retrieval_index_allowed": false,
+    "split": "test",
+    "split_hash": "84151df9e88c219b76751677558cc5bbbc19082bf358cea7330158e9e403e25f"
+  }
+}