{ "summary_schema_version": "1.0", "experiment_id": "lcot2tree_trace_pairs", "dataset_config": "opencode-reasoning2", "source_files": { "trees": "experiment_outputs/exp_lcot2tree_trace_pairs/trees_opencode-reasoning2.jsonl", "results": "experiment_outputs/exp_lcot2tree_trace_pairs/results_opencode-reasoning2.jsonl" }, "generated_at_utc": "2026-07-16T20:10:49.102844+00:00", "definitions": { "readers": [ "length-only", "format5", "gatv2" ], "primary_metric": "strict_same_group_pairwise_accuracy", "tie_tolerance": 1e-12, "gatv2_primary_aggregation": "mean_and_sample_std_across_five_training_seeds", "no_cross_problem_ranking_metric": true }, "bootstrap": { "unit": "problem_id_with_all_model_groups", "samples": 5000, "base_seed": 42, "interval": "percentile_95", "bootstrap_se_definition": "sample standard deviation of bootstrap estimates" }, "overall": { "selected_pair_count": 2000, "valid_pair_count": 1873, "invalid_pair_count": 127, "exclusion_reasons": { "negative_invalid": 70, "positive_invalid": 61 }, "construction": { "selected_groups": 2000, "trace_count": 4000, "valid_trace_count": 3869, "invalid_trace_count": 131, "valid_pair_count": 1873, "invalid_pair_count": 127, "invalid_reasons": { "assign_step[0:3]: assign_step failed after 5 attempts": 3, "assign_step[0:3]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 71, "assign_step[0:4]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 2, "assign_step[0:5]: assign_step failed after 5 attempts; last error: ValueError: Only thought B0 may have an empty assignment": 1, "assign_step[0:6]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[0:6]: assign_step failed after 5 attempts; last error: ValueError: Only thought B0 may have an empty assignment": 1, "assign_step[10:12]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[10:15]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[12:25]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[14:15]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[14:16]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[20:22]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[21:22]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[22:23]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[27:28]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[27:29]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[27:30]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[28:29]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[29:30]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[30:31]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[38:39]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[39:39]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[3:4]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[42:43]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[42:44]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[45:46]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 2, "assign_step[47:50]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[4:4]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[4:5]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[50:51]: assign_step failed after 5 attempts": 1, "assign_step[53:54]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[53:55]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[54:55]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 2, "assign_step[56:58]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[59:65]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[5:6]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[5:7]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[62:63]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[63:64]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[67:68]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[67:70]: assign_step failed after 5 attempts": 1, "assign_step[69:70]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[6:10]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[70:72]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[71:72]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 2, "assign_step[71:74]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 2, "assign_step[72:77]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[73:73]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[75:78]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[7:10]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[8:10]: assign_step failed after 5 attempts; last error: ValueError: Only thought B0 may have an empty assignment": 1, "extract_sketch: extract_sketch failed after 5 attempts": 1, "extract_sketch: extract_sketch failed after 5 attempts; last error: ValueError: Sketch contains duplicate or empty steps": 2 }, "api_response_count": 253207, "api_usage": { "prompt_tokens": 252757883, "prompt_cache_hit_tokens": 32809984, "prompt_cache_miss_tokens": 219947899, "completion_tokens": 5888802, "total_tokens": 258646685 }, "estimated_cost_cny": 232.38170268000005, "estimated_cost_per_group": { "sample_count": 2000, "sample_mean": 0.11619085134, "sample_std": 0.050956398959129345, "sample_median": 0.12084393999999998, "mean_pm_std": "0.116191 +/- 0.050956" }, "approximate_parallel_wall_seconds": 13194.415091, "tree_diagnostics": { "thought_count": { "positive": { "sample_count": 1873, "sample_mean": 47.55899626268019, "sample_std": 23.48512589508969, "sample_median": 46.0, "mean_pm_std": "47.558996 +/- 23.485126" }, "negative": { "sample_count": 1873, "sample_mean": 53.59369994660972, "sample_std": 26.930860805160467, "sample_median": 52.0, "mean_pm_std": "53.593700 +/- 26.930861" }, "difference": { "sample_count": 1873, "sample_mean": -6.034703683929525, "sample_std": 22.600638466944662, "sample_median": -4.0, "mean_pm_std": "-6.034704 +/- 22.600638" } }, "reasoning_token_count": { "positive": { "sample_count": 1873, "sample_mean": 10249.053924185799, "sample_std": 4935.609809544233, "sample_median": 10652.0, "mean_pm_std": "10249.053924 +/- 4935.609810" }, "negative": { "sample_count": 1873, "sample_mean": 11184.544046983448, "sample_std": 5236.961352297469, "sample_median": 11817.0, "mean_pm_std": "11184.544047 +/- 5236.961352" }, "difference": { "sample_count": 1873, "sample_mean": -935.4901227976509, "sample_std": 3873.0493484304866, "sample_median": -472.0, "mean_pm_std": "-935.490123 +/- 3873.049348" } }, "graph_node_count": { "positive": { "sample_count": 1873, "sample_mean": 86.30699412706888, "sample_std": 40.76859388053783, "sample_median": 84.0, "mean_pm_std": "86.306994 +/- 40.768594" }, "negative": { "sample_count": 1873, "sample_mean": 97.66631073144687, "sample_std": 46.17527715343059, "sample_median": 95.0, "mean_pm_std": "97.666311 +/- 46.175277" }, "difference": { "sample_count": 1873, "sample_mean": -11.359316604378003, "sample_std": 41.386295033824155, "sample_median": -8.0, "mean_pm_std": "-11.359317 +/- 41.386295" } }, "directed_edge_count": { "positive": { "sample_count": 1873, "sample_mean": 170.61398825413775, "sample_std": 81.53718776107566, "sample_median": 166.0, "mean_pm_std": "170.613988 +/- 81.537188" }, "negative": { "sample_count": 1873, "sample_mean": 193.33262146289374, "sample_std": 92.35055430686118, "sample_median": 188.0, "mean_pm_std": "193.332621 +/- 92.350554" }, "difference": { "sample_count": 1873, "sample_mean": -22.718633208756007, "sample_std": 82.77259006764831, "sample_median": -16.0, "mean_pm_std": "-22.718633 +/- 82.772590" } } } }, "reader_metrics": { "length-only": { "positive": { "sample_count": 1873, "sample_mean": 0.5041127258500172, "sample_std": 0.04428099836147712, "sample_median": 0.5004111464943044, "mean_pm_std": "0.504113 +/- 0.044281" }, "negative": { "sample_count": 1873, "sample_mean": 0.49578944659189245, "sample_std": 0.04692453468714944, "sample_median": 0.4899645180750672, "mean_pm_std": "0.495789 +/- 0.046925" }, "difference": { "sample_count": 1873, "sample_mean": 0.008323279258124839, "sample_std": 0.03462194777967713, "sample_median": 0.0042028420456141635, "mean_pm_std": "0.008323 +/- 0.034622" }, "ordering": { "positive_wins": 1092, "negative_wins": 781, "ties": 0, "strict_pairwise_accuracy": 0.5830218900160171, "negative_win_rate": 0.41697810998398294, "tie_rate": 0.0, "cohens_dz": 0.2404047083396779, "exact_sign_test": { "non_tied_pairs": 1873, "two_sided_p_value": 7.014206213179588e-13 }, "paired_wilcoxon": { "nonzero_pairs": 1873, "statistic": 660214.0, "two_sided_p_value": 1.662841962256126e-20 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.504121407125975, "bootstrap_se": 0.0010290025228209371, "ci95": [ 0.5021210770384934, 0.5061267901874653 ] }, "negative_mean": { "bootstrap_mean": 0.4958126915540137, "bootstrap_se": 0.0010830373955622622, "ci95": [ 0.49366124538293416, 0.4979309239336595 ] }, "mean_difference": { "bootstrap_mean": 0.008308715571961427, "bootstrap_se": 0.0008076122293290482, "ci95": [ 0.006738089073217929, 0.009914950736039969 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5828464495461826, "bootstrap_se": 0.011558787424056064, "ci95": [ 0.5600640683395622, 0.605979711692472 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "format5": { "positive": { "sample_count": 1873, "sample_mean": 0.5051087987715942, "sample_std": 0.05040684293824209, "sample_median": 0.5046998834890599, "mean_pm_std": "0.505109 +/- 0.050407" }, "negative": { "sample_count": 1873, "sample_mean": 0.49505196795641887, "sample_std": 0.052643149031888575, "sample_median": 0.49270301699546665, "mean_pm_std": "0.495052 +/- 0.052643" }, "difference": { "sample_count": 1873, "sample_mean": 0.010056830815175274, "sample_std": 0.044519273932346726, "sample_median": 0.008182633732527145, "mean_pm_std": "0.010057 +/- 0.044519" }, "ordering": { "positive_wins": 1079, "negative_wins": 794, "ties": 0, "strict_pairwise_accuracy": 0.5760811532301121, "negative_win_rate": 0.4239188467698879, "tie_rate": 0.0, "cohens_dz": 0.22589835652886067, "exact_sign_test": { "non_tied_pairs": 1873, "two_sided_p_value": 4.878951090998031e-11 }, "paired_wilcoxon": { "nonzero_pairs": 1873, "statistic": 656440.0, "two_sided_p_value": 3.614110386147969e-21 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.505119564885601, "bootstrap_se": 0.001161747253466615, "ci95": [ 0.5028805204372904, 0.5074077890670566 ] }, "negative_mean": { "bootstrap_mean": 0.4950895246484606, "bootstrap_se": 0.0011996796722296107, "ci95": [ 0.49276617752257923, 0.4975021957010411 ] }, "mean_difference": { "bootstrap_mean": 0.010030040237140388, "bootstrap_se": 0.001039382289955735, "ci95": [ 0.008012802910253925, 0.012100065613601345 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5758325680726109, "bootstrap_se": 0.011470985717364887, "ci95": [ 0.5531233315536572, 0.598505072076882 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "gatv2": { "by_seed": { "42": { "positive": { "sample_count": 1873, "sample_mean": 0.49792676926200174, "sample_std": 0.0770814723428713, "sample_median": 0.5112596750259399, "mean_pm_std": "0.497927 +/- 0.077081" }, "negative": { "sample_count": 1873, "sample_mean": 0.475221491459055, "sample_std": 0.09123596631305703, "sample_median": 0.4936100244522095, "mean_pm_std": "0.475221 +/- 0.091236" }, "difference": { "sample_count": 1873, "sample_mean": 0.02270527780294673, "sample_std": 0.08858926272242486, "sample_median": 0.014156460762023926, "mean_pm_std": "0.022705 +/- 0.088589" }, "ordering": { "positive_wins": 1105, "negative_wins": 768, "ties": 0, "strict_pairwise_accuracy": 0.589962626801922, "negative_win_rate": 0.4100373731980779, "tie_rate": 0.0, "cohens_dz": 0.2562983041645664, "exact_sign_test": { "non_tied_pairs": 1873, "two_sided_p_value": 7.001948134491203e-15 }, "paired_wilcoxon": { "nonzero_pairs": 1873, "statistic": 637679.0, "two_sided_p_value": 1.2504374829320263e-24 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.4979253200092706, "bootstrap_se": 0.0017613438986438753, "ci95": [ 0.4943880997329664, 0.5013526368762918 ] }, "negative_mean": { "bootstrap_mean": 0.4752408921389113, "bootstrap_se": 0.002085438961458828, "ci95": [ 0.4712596212941368, 0.47941227581862156 ] }, "mean_difference": { "bootstrap_mean": 0.022684427870359337, "bootstrap_se": 0.002034642898821909, "ci95": [ 0.01866348792146941, 0.026728186585223898 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5898994127068873, "bootstrap_se": 0.011401159383953838, "ci95": [ 0.5664709022957821, 0.612386545648692 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "43": { "positive": { "sample_count": 1873, "sample_mean": 0.5037567821324858, "sample_std": 0.07870015340631299, "sample_median": 0.5200029611587524, "mean_pm_std": "0.503757 +/- 0.078700" }, "negative": { "sample_count": 1873, "sample_mean": 0.48089583440063727, "sample_std": 0.09347211927011137, "sample_median": 0.5060904026031494, "mean_pm_std": "0.480896 +/- 0.093472" }, "difference": { "sample_count": 1873, "sample_mean": 0.022860947731848522, "sample_std": 0.0893780297414934, "sample_median": 0.011469125747680664, "mean_pm_std": "0.022861 +/- 0.089378" }, "ordering": { "positive_wins": 1101, "negative_wins": 772, "ties": 0, "strict_pairwise_accuracy": 0.5878270154831821, "negative_win_rate": 0.41217298451681794, "tie_rate": 0.0, "cohens_dz": 0.25577815709261953, "exact_sign_test": { "non_tied_pairs": 1873, "two_sided_p_value": 3.0045547431377115e-14 }, "paired_wilcoxon": { "nonzero_pairs": 1873, "statistic": 644180.5, "two_sided_p_value": 2.1266986730526565e-23 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5037329481847039, "bootstrap_se": 0.0018146630189806046, "ci95": [ 0.5003432988338828, 0.5072598875336671 ] }, "negative_mean": { "bootstrap_mean": 0.48090790698045777, "bootstrap_se": 0.0021614174200764165, "ci95": [ 0.47657818899704246, 0.4851438113561351 ] }, "mean_difference": { "bootstrap_mean": 0.022825041204246204, "bootstrap_se": 0.002074049931461112, "ci95": [ 0.018798474475614096, 0.02690565582718829 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5877576081153231, "bootstrap_se": 0.011432972949179705, "ci95": [ 0.5654030966364122, 0.6102509343299519 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "44": { "positive": { "sample_count": 1873, "sample_mean": 0.5229027269122694, "sample_std": 0.08431304764890894, "sample_median": 0.5228886008262634, "mean_pm_std": "0.522903 +/- 0.084313" }, "negative": { "sample_count": 1873, "sample_mean": 0.49979820504219724, "sample_std": 0.09476128428172714, "sample_median": 0.5114961862564087, "mean_pm_std": "0.499798 +/- 0.094761" }, "difference": { "sample_count": 1873, "sample_mean": 0.023104521870072104, "sample_std": 0.09133756341043993, "sample_median": 0.01209878921508789, "mean_pm_std": "0.023105 +/- 0.091338" }, "ordering": { "positive_wins": 1126, "negative_wins": 747, "ties": 0, "strict_pairwise_accuracy": 0.601174586225307, "negative_win_rate": 0.398825413774693, "tie_rate": 0.0, "cohens_dz": 0.2529575019014712, "exact_sign_test": { "non_tied_pairs": 1873, "two_sided_p_value": 1.886263686618102e-18 }, "paired_wilcoxon": { "nonzero_pairs": 1873, "statistic": 634122.0, "two_sided_p_value": 2.5687557495543967e-25 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5228874853559041, "bootstrap_se": 0.001934985061989355, "ci95": [ 0.51904397080098, 0.5266643794662197 ] }, "negative_mean": { "bootstrap_mean": 0.49982388072063894, "bootstrap_se": 0.002165131747813249, "ci95": [ 0.49553710940918994, 0.5040622544203907 ] }, "mean_difference": { "bootstrap_mean": 0.023063604635265154, "bootstrap_se": 0.0021117295070257037, "ci95": [ 0.019061175355010503, 0.02736603453441126 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.600991564335291, "bootstrap_se": 0.011425925798499324, "ci95": [ 0.5787506673785371, 0.6235985050720769 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "45": { "positive": { "sample_count": 1873, "sample_mean": 0.5195003366699463, "sample_std": 0.07178931461432175, "sample_median": 0.5301323533058167, "mean_pm_std": "0.519500 +/- 0.071789" }, "negative": { "sample_count": 1873, "sample_mean": 0.49762091451924684, "sample_std": 0.08660572989082567, "sample_median": 0.5143624544143677, "mean_pm_std": "0.497621 +/- 0.086606" }, "difference": { "sample_count": 1873, "sample_mean": 0.021879422150699455, "sample_std": 0.08871216684798479, "sample_median": 0.011796712875366211, "mean_pm_std": "0.021879 +/- 0.088712" }, "ordering": { "positive_wins": 1089, "negative_wins": 784, "ties": 0, "strict_pairwise_accuracy": 0.5814201815269621, "negative_win_rate": 0.41857981847303793, "tie_rate": 0.0, "cohens_dz": 0.24663383759063795, "exact_sign_test": { "non_tied_pairs": 1873, "two_sided_p_value": 1.9282040822544803e-12 }, "paired_wilcoxon": { "nonzero_pairs": 1873, "statistic": 656669.0, "two_sided_p_value": 3.967713200218313e-21 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5194778371205224, "bootstrap_se": 0.0016246929162932917, "ci95": [ 0.5163405365432432, 0.5226653068907682 ] }, "negative_mean": { "bootstrap_mean": 0.49764261810222515, "bootstrap_se": 0.0019959826332099054, "ci95": [ 0.4937303319358597, 0.5016303620697483 ] }, "mean_difference": { "bootstrap_mean": 0.02183521901829734, "bootstrap_se": 0.0020686169991017715, "ci95": [ 0.01776504711756653, 0.025940005877440584 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5813160704751735, "bootstrap_se": 0.011375275141641284, "ci95": [ 0.5595168179391351, 0.604378003203417 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "46": { "positive": { "sample_count": 1873, "sample_mean": 0.52020099600782, "sample_std": 0.06281305726781251, "sample_median": 0.5258979797363281, "mean_pm_std": "0.520201 +/- 0.062813" }, "negative": { "sample_count": 1873, "sample_mean": 0.5023361757373784, "sample_std": 0.07300595001032742, "sample_median": 0.512582540512085, "mean_pm_std": "0.502336 +/- 0.073006" }, "difference": { "sample_count": 1873, "sample_mean": 0.01786482027044159, "sample_std": 0.07478553551078312, "sample_median": 0.01062101125717163, "mean_pm_std": "0.017865 +/- 0.074786" }, "ordering": { "positive_wins": 1079, "negative_wins": 794, "ties": 0, "strict_pairwise_accuracy": 0.5760811532301121, "negative_win_rate": 0.4239188467698879, "tie_rate": 0.0, "cohens_dz": 0.2388806892726697, "exact_sign_test": { "non_tied_pairs": 1873, "two_sided_p_value": 4.878951090998031e-11 }, "paired_wilcoxon": { "nonzero_pairs": 1873, "statistic": 660154.0, "two_sided_p_value": 1.6233051800784045e-20 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5201845406994217, "bootstrap_se": 0.0014379384408963782, "ci95": [ 0.5173613356351773, 0.522989135963899 ] }, "negative_mean": { "bootstrap_mean": 0.5023477095303851, "bootstrap_se": 0.0016753068274928712, "ci95": [ 0.4990712073820037, 0.5056399353357699 ] }, "mean_difference": { "bootstrap_mean": 0.01783683116903645, "bootstrap_se": 0.001714898195896716, "ci95": [ 0.014513989350794792, 0.021146614491955453 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5759798184730379, "bootstrap_se": 0.01149183188905047, "ci95": [ 0.5531233315536572, 0.598505072076882 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } } }, "strict_pairwise_accuracy_across_seeds": { "sample_count": 5, "sample_mean": 0.5872931126534969, "sample_std": 0.009490864300390367, "sample_median": 0.5878270154831821, "mean_pm_std": "0.587293 +/- 0.009491" }, "bootstrap_of_seed_mean_strict_pairwise_accuracy": { "bootstrap_mean": 0.5871888948211426, "bootstrap_se": 0.00917479969819558, "ci95": [ 0.5688200747463963, 0.604911906033102 ] } } }, "gatv2_seed_variability": { "seeds": [ 42, 43, 44, 45, 46 ], "metric": "strict_pairwise_accuracy", "values": [ 0.589962626801922, 0.5878270154831821, 0.601174586225307, 0.5814201815269621, 0.5760811532301121 ], "mean_pm_std": "0.587293 +/- 0.009491" }, "paired_incremental_validity": { "gatv2_minus_length-only": { "strict_pairwise_accuracy_difference": 0.00427122263747981, "bootstrap": { "bootstrap_mean": 0.004342445274959955, "bootstrap_se": 0.011779776359026627, "ci95": [ -0.018259476775226883, 0.027549386011745728 ] } }, "gatv2_minus_format5": { "strict_pairwise_accuracy_difference": 0.01121195942338482, "bootstrap": { "bootstrap_mean": 0.011356326748531766, "bootstrap_se": 0.012816657170404654, "ci95": [ -0.01356380138814741, 0.03683929524826479 ] } } } }, "by_source_subset": { "cpp": { "selected_pair_count": 409, "valid_pair_count": 378, "invalid_pair_count": 31, "exclusion_reasons": { "negative_invalid": 19, "positive_invalid": 13 }, "construction": { "selected_groups": 409, "trace_count": 818, "valid_trace_count": 786, "invalid_trace_count": 32, "valid_pair_count": 378, "invalid_pair_count": 31, "invalid_reasons": { "assign_step[0:3]: assign_step failed after 5 attempts": 1, "assign_step[0:3]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 15, "assign_step[0:5]: assign_step failed after 5 attempts; last error: ValueError: Only thought B0 may have an empty assignment": 1, "assign_step[21:22]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[27:28]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[27:29]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[27:30]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[29:30]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[30:31]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[39:39]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[45:46]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[47:50]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[53:54]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[54:55]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[5:7]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[63:64]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[69:70]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[6:10]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1 }, "api_response_count": 51113, "api_usage": { "prompt_tokens": 51196784, "prompt_cache_hit_tokens": 6664960, "prompt_cache_miss_tokens": 44531824, "completion_tokens": 1188930, "total_tokens": 52385714 }, "estimated_cost_cny": 47.04298320000003, "estimated_cost_per_group": { "sample_count": 409, "sample_mean": 0.11501951882640586, "sample_std": 0.03902980742661729, "sample_median": 0.12301196, "mean_pm_std": "0.115020 +/- 0.039030" }, "approximate_parallel_wall_seconds": 13187.947359, "tree_diagnostics": { "thought_count": { "positive": { "sample_count": 378, "sample_mean": 47.24867724867725, "sample_std": 21.424385839284064, "sample_median": 46.0, "mean_pm_std": "47.248677 +/- 21.424386" }, "negative": { "sample_count": 378, "sample_mean": 52.84126984126984, "sample_std": 25.75957374962824, "sample_median": 49.0, "mean_pm_std": "52.841270 +/- 25.759574" }, "difference": { "sample_count": 378, "sample_mean": -5.592592592592593, "sample_std": 22.913009246498607, "sample_median": -3.5, "mean_pm_std": "-5.592593 +/- 22.913009" } }, "reasoning_token_count": { "positive": { "sample_count": 378, "sample_mean": 10166.457671957673, "sample_std": 3871.5383120132765, "sample_median": 10941.0, "mean_pm_std": "10166.457672 +/- 3871.538312" }, "negative": { "sample_count": 378, "sample_mean": 11191.952380952382, "sample_std": 3915.828218862036, "sample_median": 12306.0, "mean_pm_std": "11191.952381 +/- 3915.828219" }, "difference": { "sample_count": 378, "sample_mean": -1025.494708994709, "sample_std": 3238.8464579161537, "sample_median": -573.5, "mean_pm_std": "-1025.494709 +/- 3238.846458" } }, "graph_node_count": { "positive": { "sample_count": 378, "sample_mean": 82.83333333333333, "sample_std": 33.905254946650196, "sample_median": 81.0, "mean_pm_std": "82.833333 +/- 33.905255" }, "negative": { "sample_count": 378, "sample_mean": 95.02645502645503, "sample_std": 40.3463695640878, "sample_median": 95.0, "mean_pm_std": "95.026455 +/- 40.346370" }, "difference": { "sample_count": 378, "sample_mean": -12.193121693121693, "sample_std": 38.24432490679713, "sample_median": -8.0, "mean_pm_std": "-12.193122 +/- 38.244325" } }, "directed_edge_count": { "positive": { "sample_count": 378, "sample_mean": 163.66666666666666, "sample_std": 67.81050989330039, "sample_median": 160.0, "mean_pm_std": "163.666667 +/- 67.810510" }, "negative": { "sample_count": 378, "sample_mean": 188.05291005291005, "sample_std": 80.6927391281756, "sample_median": 188.0, "mean_pm_std": "188.052910 +/- 80.692739" }, "difference": { "sample_count": 378, "sample_mean": -24.386243386243386, "sample_std": 76.48864981359426, "sample_median": -16.0, "mean_pm_std": "-24.386243 +/- 76.488650" } } } }, "reader_metrics": { "length-only": { "positive": { "sample_count": 378, "sample_mean": 0.5049142784745002, "sample_std": 0.034813381634429576, "sample_median": 0.4979252593228507, "mean_pm_std": "0.504914 +/- 0.034813" }, "negative": { "sample_count": 378, "sample_mean": 0.49570626825810077, "sample_std": 0.035273886825100326, "sample_median": 0.48584163813895587, "mean_pm_std": "0.495706 +/- 0.035274" }, "difference": { "sample_count": 378, "sample_mean": 0.009208010216399444, "sample_std": 0.029082439050962677, "sample_median": 0.0052885619825065455, "mean_pm_std": "0.009208 +/- 0.029082" }, "ordering": { "positive_wins": 229, "negative_wins": 149, "ties": 0, "strict_pairwise_accuracy": 0.6058201058201058, "negative_win_rate": 0.3941798941798942, "tie_rate": 0.0, "cohens_dz": 0.3166175367981952, "exact_sign_test": { "non_tied_pairs": 378, "two_sided_p_value": 4.5534313310350145e-05 }, "paired_wilcoxon": { "nonzero_pairs": 378, "statistic": 24362.0, "two_sided_p_value": 7.123224086372473e-08 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5049300962188631, "bootstrap_se": 0.0017691764385072055, "ci95": [ 0.5014450538992209, 0.5083839162175458 ] }, "negative_mean": { "bootstrap_mean": 0.4957223285132037, "bootstrap_se": 0.0018205025750961349, "ci95": [ 0.4921728367427629, 0.4992279771317377 ] }, "mean_difference": { "bootstrap_mean": 0.009207767705659353, "bootstrap_se": 0.001501441536670455, "ci95": [ 0.006365878173399752, 0.012242805109542768 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.6057608465608466, "bootstrap_se": 0.025142070259471752, "ci95": [ 0.5555555555555556, 0.656084656084656 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "format5": { "positive": { "sample_count": 378, "sample_mean": 0.4979617543604543, "sample_std": 0.04206935422670228, "sample_median": 0.4985154605133407, "mean_pm_std": "0.497962 +/- 0.042069" }, "negative": { "sample_count": 378, "sample_mean": 0.4865417601940175, "sample_std": 0.04241782574516262, "sample_median": 0.4842078826728945, "mean_pm_std": "0.486542 +/- 0.042418" }, "difference": { "sample_count": 378, "sample_mean": 0.011419994166436884, "sample_std": 0.041985825529314125, "sample_median": 0.009064886962245139, "mean_pm_std": "0.011420 +/- 0.041986" }, "ordering": { "positive_wins": 227, "negative_wins": 151, "ties": 0, "strict_pairwise_accuracy": 0.6005291005291006, "negative_win_rate": 0.3994708994708995, "tie_rate": 0.0, "cohens_dz": 0.2719964183737092, "exact_sign_test": { "non_tied_pairs": 378, "two_sided_p_value": 0.00010903214719687808 }, "paired_wilcoxon": { "nonzero_pairs": 378, "statistic": 24660.0, "two_sided_p_value": 1.5388648911590643e-07 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.49793344784276294, "bootstrap_se": 0.002130423878069571, "ci95": [ 0.49372704659077676, 0.5021182182498014 ] }, "negative_mean": { "bootstrap_mean": 0.48653335056425806, "bootstrap_se": 0.0021707028825402866, "ci95": [ 0.4822928467744062, 0.49071010854377745 ] }, "mean_difference": { "bootstrap_mean": 0.011400097278504866, "bootstrap_se": 0.002177250121003969, "ci95": [ 0.007065097265410799, 0.015719217183702793 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.6004857142857143, "bootstrap_se": 0.025465716634259217, "ci95": [ 0.5502645502645502, 0.6482142857142847 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "gatv2": { "by_seed": { "42": { "positive": { "sample_count": 378, "sample_mean": 0.5065981955834167, "sample_std": 0.0696020945627947, "sample_median": 0.5143698453903198, "mean_pm_std": "0.506598 +/- 0.069602" }, "negative": { "sample_count": 378, "sample_mean": 0.4843486700719469, "sample_std": 0.08329181772497005, "sample_median": 0.4977763742208481, "mean_pm_std": "0.484349 +/- 0.083292" }, "difference": { "sample_count": 378, "sample_mean": 0.02224952551146979, "sample_std": 0.08729675830040941, "sample_median": 0.013928890228271484, "mean_pm_std": "0.022250 +/- 0.087297" }, "ordering": { "positive_wins": 227, "negative_wins": 151, "ties": 0, "strict_pairwise_accuracy": 0.6005291005291006, "negative_win_rate": 0.3994708994708995, "tie_rate": 0.0, "cohens_dz": 0.254872299323003, "exact_sign_test": { "non_tied_pairs": 378, "two_sided_p_value": 0.00010903214719687808 }, "paired_wilcoxon": { "nonzero_pairs": 378, "statistic": 25576.0, "two_sided_p_value": 1.4577901193699682e-06 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5066915041223089, "bootstrap_se": 0.003628567768162753, "ci95": [ 0.4994222690168985, 0.5134381664079175 ] }, "negative_mean": { "bootstrap_mean": 0.48434349121259, "bootstrap_se": 0.00428957635038576, "ci95": [ 0.47599860159113566, 0.4924819220234418 ] }, "mean_difference": { "bootstrap_mean": 0.022348012909718922, "bootstrap_se": 0.004538403481871207, "ci95": [ 0.013590624729890831, 0.031315324285742685 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.6017185185185187, "bootstrap_se": 0.02506264511443439, "ci95": [ 0.5529100529100529, 0.6507936507936508 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "43": { "positive": { "sample_count": 378, "sample_mean": 0.5174519489051173, "sample_std": 0.06901087316313649, "sample_median": 0.5257491171360016, "mean_pm_std": "0.517452 +/- 0.069011" }, "negative": { "sample_count": 378, "sample_mean": 0.4959224732937636, "sample_std": 0.08667945355764867, "sample_median": 0.5101312696933746, "mean_pm_std": "0.495922 +/- 0.086679" }, "difference": { "sample_count": 378, "sample_mean": 0.02152947561135368, "sample_std": 0.08943207380514161, "sample_median": 0.012146860361099243, "mean_pm_std": "0.021529 +/- 0.089432" }, "ordering": { "positive_wins": 225, "negative_wins": 153, "ties": 0, "strict_pairwise_accuracy": 0.5952380952380952, "negative_win_rate": 0.40476190476190477, "tie_rate": 0.0, "cohens_dz": 0.24073550679662212, "exact_sign_test": { "non_tied_pairs": 378, "two_sided_p_value": 0.00025034595398356165 }, "paired_wilcoxon": { "nonzero_pairs": 378, "statistic": 26966.0, "two_sided_p_value": 3.140340276756643e-05 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5174727475075338, "bootstrap_se": 0.003526438913067313, "ci95": [ 0.5107060729254964, 0.5244148810645417 ] }, "negative_mean": { "bootstrap_mean": 0.4958596115992025, "bootstrap_se": 0.004500419642715377, "ci95": [ 0.48683732381415745, 0.5044285157261781 ] }, "mean_difference": { "bootstrap_mean": 0.021613135908331187, "bootstrap_se": 0.004652942225522844, "ci95": [ 0.012636569098978446, 0.030761860342568184 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5955931216931217, "bootstrap_se": 0.02503888335052725, "ci95": [ 0.5476190476190477, 0.6455026455026455 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "44": { "positive": { "sample_count": 378, "sample_mean": 0.5290711731033981, "sample_std": 0.07437263394532395, "sample_median": 0.5245527625083923, "mean_pm_std": "0.529071 +/- 0.074373" }, "negative": { "sample_count": 378, "sample_mean": 0.5082962466847329, "sample_std": 0.08819658250774336, "sample_median": 0.508731484413147, "mean_pm_std": "0.508296 +/- 0.088197" }, "difference": { "sample_count": 378, "sample_mean": 0.020774926418665227, "sample_std": 0.0919766076552159, "sample_median": 0.0141192227602005, "mean_pm_std": "0.020775 +/- 0.091977" }, "ordering": { "positive_wins": 226, "negative_wins": 152, "ties": 0, "strict_pairwise_accuracy": 0.5978835978835979, "negative_win_rate": 0.4021164021164021, "tie_rate": 0.0, "cohens_dz": 0.22587184881336622, "exact_sign_test": { "non_tied_pairs": 378, "two_sided_p_value": 0.00016607967341947117 }, "paired_wilcoxon": { "nonzero_pairs": 378, "statistic": 26162.0, "two_sided_p_value": 5.591732647804814e-06 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5291437521519169, "bootstrap_se": 0.003758736036193227, "ci95": [ 0.5218218705089634, 0.536724568785183 ] }, "negative_mean": { "bootstrap_mean": 0.5082640321533358, "bootstrap_se": 0.004555865110199071, "ci95": [ 0.49935668305094755, 0.5172315624419344 ] }, "mean_difference": { "bootstrap_mean": 0.02087971999858107, "bootstrap_se": 0.004748255897213782, "ci95": [ 0.011643771827220916, 0.030112789189886455 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5986306878306878, "bootstrap_se": 0.025297000909213295, "ci95": [ 0.5502645502645502, 0.6481481481481481 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "45": { "positive": { "sample_count": 378, "sample_mean": 0.5323515453981975, "sample_std": 0.06425083055038089, "sample_median": 0.5359054207801819, "mean_pm_std": "0.532352 +/- 0.064251" }, "negative": { "sample_count": 378, "sample_mean": 0.508478238823868, "sample_std": 0.07862082382387173, "sample_median": 0.5177236199378967, "mean_pm_std": "0.508478 +/- 0.078621" }, "difference": { "sample_count": 378, "sample_mean": 0.023873306574329498, "sample_std": 0.08944558071158704, "sample_median": 0.020120769739151, "mean_pm_std": "0.023873 +/- 0.089446" }, "ordering": { "positive_wins": 233, "negative_wins": 145, "ties": 0, "strict_pairwise_accuracy": 0.6164021164021164, "negative_win_rate": 0.3835978835978836, "tie_rate": 0.0, "cohens_dz": 0.26690314249630537, "exact_sign_test": { "non_tied_pairs": 378, "two_sided_p_value": 6.991951892200577e-06 }, "paired_wilcoxon": { "nonzero_pairs": 378, "statistic": 25359.0, "two_sided_p_value": 8.697893706175397e-07 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5323849580510741, "bootstrap_se": 0.0033414825955238466, "ci95": [ 0.5258941269248093, 0.5389052013142241 ] }, "negative_mean": { "bootstrap_mean": 0.5084404704575185, "bootstrap_se": 0.004079940488426655, "ci95": [ 0.5003664244813893, 0.5163388103690175 ] }, "mean_difference": { "bootstrap_mean": 0.02394448759355558, "bootstrap_se": 0.0046780851004316224, "ci95": [ 0.014807919231514451, 0.0331558323874281 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.6169619047619048, "bootstrap_se": 0.025365226650048242, "ci95": [ 0.5687830687830688, 0.6666666666666666 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "46": { "positive": { "sample_count": 378, "sample_mean": 0.5333233809187299, "sample_std": 0.05501341194326524, "sample_median": 0.532279908657074, "mean_pm_std": "0.533323 +/- 0.055013" }, "negative": { "sample_count": 378, "sample_mean": 0.5138634355374114, "sample_std": 0.06377177522456245, "sample_median": 0.5180603563785553, "mean_pm_std": "0.513863 +/- 0.063772" }, "difference": { "sample_count": 378, "sample_mean": 0.019459945381318452, "sample_std": 0.07044951505348193, "sample_median": 0.012625366449356079, "mean_pm_std": "0.019460 +/- 0.070450" }, "ordering": { "positive_wins": 226, "negative_wins": 152, "ties": 0, "strict_pairwise_accuracy": 0.5978835978835979, "negative_win_rate": 0.4021164021164021, "tie_rate": 0.0, "cohens_dz": 0.27622539866378615, "exact_sign_test": { "non_tied_pairs": 378, "two_sided_p_value": 0.00016607967341947117 }, "paired_wilcoxon": { "nonzero_pairs": 378, "statistic": 25397.0, "two_sided_p_value": 9.528067718856672e-07 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.533335671973733, "bootstrap_se": 0.0028507922374497625, "ci95": [ 0.5277062349001724, 0.5388833587644276 ] }, "negative_mean": { "bootstrap_mean": 0.513844784401854, "bootstrap_se": 0.003313263857806373, "ci95": [ 0.5073324129419037, 0.5202331037630165 ] }, "mean_difference": { "bootstrap_mean": 0.019490887571879164, "bootstrap_se": 0.0036737148521993122, "ci95": [ 0.012299768920376819, 0.026522189442797623 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5984174603174603, "bootstrap_se": 0.02518840952992547, "ci95": [ 0.5476190476190477, 0.6481481481481481 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } } }, "strict_pairwise_accuracy_across_seeds": { "sample_count": 5, "sample_mean": 0.6015873015873016, "sample_std": 0.00849037382850857, "sample_median": 0.5978835978835979, "mean_pm_std": "0.601587 +/- 0.008490" }, "bootstrap_of_seed_mean_strict_pairwise_accuracy": { "bootstrap_mean": 0.6022643386243386, "bootstrap_se": 0.02016732442887762, "ci95": [ 0.5634920634920635, 0.6423280423280423 ] } } }, "gatv2_seed_variability": { "seeds": [ 42, 43, 44, 45, 46 ], "metric": "strict_pairwise_accuracy", "values": [ 0.6005291005291006, 0.5952380952380952, 0.5978835978835979, 0.6164021164021164, 0.5978835978835979 ], "mean_pm_std": "0.601587 +/- 0.008490" }, "paired_incremental_validity": { "gatv2_minus_length-only": { "strict_pairwise_accuracy_difference": -0.004232804232804255, "bootstrap": { "bootstrap_mean": -0.003496507936507936, "bootstrap_se": 0.026101194431045662, "ci95": [ -0.053981481481481394, 0.04816137566137548 ] } }, "gatv2_minus_format5": { "strict_pairwise_accuracy_difference": 0.0010582010582009804, "bootstrap": { "bootstrap_mean": 0.0017786243386243377, "bootstrap_se": 0.03043511437642681, "ci95": [ -0.058730158730158744, 0.0624338624338624 ] } } } }, "python": { "selected_pair_count": 1591, "valid_pair_count": 1495, "invalid_pair_count": 96, "exclusion_reasons": { "negative_invalid": 51, "positive_invalid": 48 }, "construction": { "selected_groups": 1591, "trace_count": 3182, "valid_trace_count": 3083, "invalid_trace_count": 99, "valid_pair_count": 1495, "invalid_pair_count": 96, "invalid_reasons": { "assign_step[0:3]: assign_step failed after 5 attempts": 2, "assign_step[0:3]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 56, "assign_step[0:4]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 2, "assign_step[0:6]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[0:6]: assign_step failed after 5 attempts; last error: ValueError: Only thought B0 may have an empty assignment": 1, "assign_step[10:12]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[10:15]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[12:25]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[14:15]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[14:16]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[20:22]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[22:23]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[28:29]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[38:39]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[3:4]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[42:43]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[42:44]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[45:46]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[4:4]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[4:5]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[50:51]: assign_step failed after 5 attempts": 1, "assign_step[53:55]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[54:55]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[56:58]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[59:65]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[5:6]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[62:63]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[67:68]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[67:70]: assign_step failed after 5 attempts": 1, "assign_step[70:72]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[71:72]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 2, "assign_step[71:74]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 2, "assign_step[72:77]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[73:73]: assign_step failed after 5 attempts; last error: ValueError: Assigned steps must be unique and ordered": 1, "assign_step[75:78]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[7:10]: assign_step failed after 5 attempts; last error: ValueError: Assignment response does not cover its complete thought chunk": 1, "assign_step[8:10]: assign_step failed after 5 attempts; last error: ValueError: Only thought B0 may have an empty assignment": 1, "extract_sketch: extract_sketch failed after 5 attempts": 1, "extract_sketch: extract_sketch failed after 5 attempts; last error: ValueError: Sketch contains duplicate or empty steps": 2 }, "api_response_count": 202094, "api_usage": { "prompt_tokens": 201561099, "prompt_cache_hit_tokens": 26145024, "prompt_cache_miss_tokens": 175416075, "completion_tokens": 4699872, "total_tokens": 206260971 }, "estimated_cost_cny": 185.3387194800002, "estimated_cost_per_group": { "sample_count": 1591, "sample_mean": 0.1164919669893149, "sample_std": 0.05360166539714576, "sample_median": 0.12002356, "mean_pm_std": "0.116492 +/- 0.053602" }, "approximate_parallel_wall_seconds": 13194.415091, "tree_diagnostics": { "thought_count": { "positive": { "sample_count": 1495, "sample_mean": 47.63745819397993, "sample_std": 23.98423040513491, "sample_median": 46.0, "mean_pm_std": "47.637458 +/- 23.984230" }, "negative": { "sample_count": 1495, "sample_mean": 53.78394648829431, "sample_std": 27.224083935521612, "sample_median": 52.0, "mean_pm_std": "53.783946 +/- 27.224084" }, "difference": { "sample_count": 1495, "sample_mean": -6.1464882943143815, "sample_std": 22.52734347837016, "sample_median": -4.0, "mean_pm_std": "-6.146488 +/- 22.527343" } }, "reasoning_token_count": { "positive": { "sample_count": 1495, "sample_mean": 10269.93779264214, "sample_std": 5170.9960730688945, "sample_median": 10552.0, "mean_pm_std": "10269.937793 +/- 5170.996073" }, "negative": { "sample_count": 1495, "sample_mean": 11182.670903010034, "sample_std": 5522.268154996848, "sample_median": 11658.0, "mean_pm_std": "11182.670903 +/- 5522.268155" }, "difference": { "sample_count": 1495, "sample_mean": -912.733110367893, "sample_std": 4018.2268029859847, "sample_median": -420.0, "mean_pm_std": "-912.733110 +/- 4018.226803" } }, "graph_node_count": { "positive": { "sample_count": 1495, "sample_mean": 87.18528428093646, "sample_std": 42.29296114507862, "sample_median": 85.0, "mean_pm_std": "87.185284 +/- 42.292961" }, "negative": { "sample_count": 1495, "sample_mean": 98.33377926421404, "sample_std": 47.52512506891524, "sample_median": 95.0, "mean_pm_std": "98.333779 +/- 47.525125" }, "difference": { "sample_count": 1495, "sample_mean": -11.148494983277592, "sample_std": 42.15313634464903, "sample_median": -8.0, "mean_pm_std": "-11.148495 +/- 42.153136" } }, "directed_edge_count": { "positive": { "sample_count": 1495, "sample_mean": 172.37056856187291, "sample_std": 84.58592229015724, "sample_median": 168.0, "mean_pm_std": "172.370569 +/- 84.585922" }, "negative": { "sample_count": 1495, "sample_mean": 194.6675585284281, "sample_std": 95.05025013783047, "sample_median": 188.0, "mean_pm_std": "194.667559 +/- 95.050250" }, "difference": { "sample_count": 1495, "sample_mean": -22.296989966555184, "sample_std": 84.30627268929805, "sample_median": -16.0, "mean_pm_std": "-22.296990 +/- 84.306273" } } } }, "reader_metrics": { "length-only": { "positive": { "sample_count": 1495, "sample_mean": 0.5039100590325895, "sample_std": 0.04637756710241164, "sample_median": 0.5013203553365156, "mean_pm_std": "0.503910 +/- 0.046378" }, "negative": { "sample_count": 1495, "sample_mean": 0.4958104776354866, "sample_std": 0.04944737951167473, "sample_median": 0.49184155417249864, "mean_pm_std": "0.495810 +/- 0.049447" }, "difference": { "sample_count": 1495, "sample_mean": 0.008099581397102897, "sample_std": 0.035892658702259495, "sample_median": 0.003812853892211643, "mean_pm_std": "0.008100 +/- 0.035893" }, "ordering": { "positive_wins": 863, "negative_wins": 632, "ties": 0, "strict_pairwise_accuracy": 0.577257525083612, "negative_win_rate": 0.42274247491638794, "tie_rate": 0.0, "cohens_dz": 0.2256612268344729, "exact_sign_test": { "non_tied_pairs": 1495, "two_sided_p_value": 2.5224314798141916e-09 }, "paired_wilcoxon": { "nonzero_pairs": 1495, "statistic": 430150.0, "two_sided_p_value": 1.113094294419744e-14 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5039195364141006, "bootstrap_se": 0.00120500690327834, "ci95": [ 0.5016076489435631, 0.506258301653583 ] }, "negative_mean": { "bootstrap_mean": 0.49579046943825067, "bootstrap_se": 0.0012905120982708027, "ci95": [ 0.49333304488907465, 0.49833274283126267 ] }, "mean_difference": { "bootstrap_mean": 0.008129066975849863, "bootstrap_se": 0.0009225081238419826, "ci95": [ 0.006338981304532805, 0.009977444764589028 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5776004013377926, "bootstrap_se": 0.012672823455458256, "ci95": [ 0.5525083612040134, 0.6026755852842809 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "format5": { "positive": { "sample_count": 1495, "sample_mean": 0.50691587755916, "sample_std": 0.05216243824393153, "sample_median": 0.5060804224064134, "mean_pm_std": "0.506916 +/- 0.052162" }, "negative": { "sample_count": 1495, "sample_mean": 0.4972037127953405, "sample_std": 0.054731009611205264, "sample_median": 0.4957549931369375, "mean_pm_std": "0.497204 +/- 0.054731" }, "difference": { "sample_count": 1495, "sample_mean": 0.009712164763819496, "sample_std": 0.045144277746951136, "sample_median": 0.007846700648831528, "mean_pm_std": "0.009712 +/- 0.045144" }, "ordering": { "positive_wins": 852, "negative_wins": 643, "ties": 0, "strict_pairwise_accuracy": 0.5698996655518395, "negative_win_rate": 0.43010033444816054, "tie_rate": 0.0, "cohens_dz": 0.2151361202024196, "exact_sign_test": { "non_tied_pairs": 1495, "two_sided_p_value": 7.126696371696971e-08 }, "paired_wilcoxon": { "nonzero_pairs": 1495, "statistic": 425993.0, "two_sided_p_value": 1.528551348707579e-15 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5069195393661136, "bootstrap_se": 0.0013673291484691493, "ci95": [ 0.5042367471049969, 0.5096109719811156 ] }, "negative_mean": { "bootstrap_mean": 0.4971847214328036, "bootstrap_se": 0.0014480105275802035, "ci95": [ 0.49430912484255585, 0.5000783618900133 ] }, "mean_difference": { "bootstrap_mean": 0.009734817933309969, "bootstrap_se": 0.0011591926210953924, "ci95": [ 0.007411440003609777, 0.011977025612706358 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5699967892976588, "bootstrap_se": 0.012960911938847883, "ci95": [ 0.5444816053511705, 0.5953177257525084 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "gatv2": { "by_seed": { "42": { "positive": { "sample_count": 1495, "sample_mean": 0.49573426146969746, "sample_std": 0.07873081198707631, "sample_median": 0.5100011229515076, "mean_pm_std": "0.495734 +/- 0.078731" }, "negative": { "sample_count": 1495, "sample_mean": 0.47291374997699936, "sample_std": 0.09302168257706582, "sample_median": 0.4926658570766449, "mean_pm_std": "0.472914 +/- 0.093022" }, "difference": { "sample_count": 1495, "sample_mean": 0.02282051149269809, "sample_std": 0.08894161354159026, "sample_median": 0.014183759689331055, "mean_pm_std": "0.022821 +/- 0.088942" }, "ordering": { "positive_wins": 878, "negative_wins": 617, "ties": 0, "strict_pairwise_accuracy": 0.5872909698996656, "negative_win_rate": 0.41270903010033444, "tie_rate": 0.0, "cohens_dz": 0.2565785641164124, "exact_sign_test": { "non_tied_pairs": 1495, "two_sided_p_value": 1.5714306272526418e-11 }, "paired_wilcoxon": { "nonzero_pairs": 1495, "statistic": 407991.0, "two_sided_p_value": 1.3922496250304106e-19 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.49575732951508916, "bootstrap_se": 0.0020254543889485976, "ci95": [ 0.4917856155171442, 0.49964235825981185 ] }, "negative_mean": { "bootstrap_mean": 0.47292274889357616, "bootstrap_se": 0.0024080579466141086, "ci95": [ 0.4682929387267517, 0.4775885621485079 ] }, "mean_difference": { "bootstrap_mean": 0.02283458062151304, "bootstrap_se": 0.002333301421773096, "ci95": [ 0.01837767946219265, 0.027382436911186147 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.587310635451505, "bootstrap_se": 0.012726868988290774, "ci95": [ 0.5632107023411371, 0.6120401337792643 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "43": { "positive": { "sample_count": 1495, "sample_mean": 0.5002940576909777, "sample_std": 0.0806198694339098, "sample_median": 0.5188823342323303, "mean_pm_std": "0.500294 +/- 0.080620" }, "negative": { "sample_count": 1495, "sample_mean": 0.47709645680759266, "sample_std": 0.09476341216504032, "sample_median": 0.5051302909851074, "mean_pm_std": "0.477096 +/- 0.094763" }, "difference": { "sample_count": 1495, "sample_mean": 0.02319760088338501, "sample_std": 0.08939115565473223, "sample_median": 0.011326730251312256, "mean_pm_std": "0.023198 +/- 0.089391" }, "ordering": { "positive_wins": 876, "negative_wins": 619, "ties": 0, "strict_pairwise_accuracy": 0.585953177257525, "negative_win_rate": 0.4140468227424749, "tie_rate": 0.0, "cohens_dz": 0.259506667225383, "exact_sign_test": { "non_tied_pairs": 1495, "two_sided_p_value": 3.203168787600772e-11 }, "paired_wilcoxon": { "nonzero_pairs": 1495, "statistic": 407808.0, "two_sided_p_value": 1.2591627813434933e-19 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5003055594032664, "bootstrap_se": 0.0021306492766813064, "ci95": [ 0.49611997661657353, 0.5043904683497239 ] }, "negative_mean": { "bootstrap_mean": 0.47707554032230454, "bootstrap_se": 0.002442689508569003, "ci95": [ 0.47227754739141187, 0.48190945725601453 ] }, "mean_difference": { "bootstrap_mean": 0.02323001908096183, "bootstrap_se": 0.0023154208206135726, "ci95": [ 0.01870238606958186, 0.027781815217265915 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5861819397993311, "bootstrap_se": 0.012672929300639032, "ci95": [ 0.5618729096989966, 0.6107023411371237 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "44": { "positive": { "sample_count": 1495, "sample_mean": 0.5213430796478903, "sample_std": 0.08659926507346592, "sample_median": 0.5226179361343384, "mean_pm_std": "0.521343 +/- 0.086599" }, "negative": { "sample_count": 1495, "sample_mean": 0.4976495363192016, "sample_std": 0.09625954310008666, "sample_median": 0.5121106505393982, "mean_pm_std": "0.497650 +/- 0.096260" }, "difference": { "sample_count": 1495, "sample_mean": 0.023693543328688695, "sample_std": 0.09119678362170007, "sample_median": 0.012006044387817383, "mean_pm_std": "0.023694 +/- 0.091197" }, "ordering": { "positive_wins": 900, "negative_wins": 595, "ties": 0, "strict_pairwise_accuracy": 0.6020066889632107, "negative_win_rate": 0.3979933110367893, "tie_rate": 0.0, "cohens_dz": 0.25980678690351167, "exact_sign_test": { "non_tied_pairs": 1495, "two_sided_p_value": 3.03594159633112e-15 }, "paired_wilcoxon": { "nonzero_pairs": 1495, "statistic": 403025.0, "two_sided_p_value": 8.735874485022314e-21 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5213607534821835, "bootstrap_se": 0.002245441937839786, "ci95": [ 0.5169353928947729, 0.5255967396789272 ] }, "negative_mean": { "bootstrap_mean": 0.49763647785356313, "bootstrap_se": 0.0024824743142325594, "ci95": [ 0.4928160888350329, 0.5024895965147227 ] }, "mean_difference": { "bootstrap_mean": 0.023724275628620364, "bootstrap_se": 0.0023719625903068984, "ci95": [ 0.019238102965081817, 0.028456002247198967 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.6022078929765886, "bootstrap_se": 0.012645753446679195, "ci95": [ 0.577257525083612, 0.6267725752508359 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "45": { "positive": { "sample_count": 1495, "sample_mean": 0.5162510009513651, "sample_std": 0.07323654039150522, "sample_median": 0.5283108353614807, "mean_pm_std": "0.516251 +/- 0.073237" }, "negative": { "sample_count": 1495, "sample_mean": 0.49487571813988446, "sample_std": 0.08832398720543459, "sample_median": 0.5136783719062805, "mean_pm_std": "0.494876 +/- 0.088324" }, "difference": { "sample_count": 1495, "sample_mean": 0.02137528281148062, "sample_std": 0.08854876646247654, "sample_median": 0.010409682989120483, "mean_pm_std": "0.021375 +/- 0.088549" }, "ordering": { "positive_wins": 856, "negative_wins": 639, "ties": 0, "strict_pairwise_accuracy": 0.5725752508361204, "negative_win_rate": 0.4274247491638796, "tie_rate": 0.0, "cohens_dz": 0.24139560228135554, "exact_sign_test": { "non_tied_pairs": 1495, "two_sided_p_value": 2.194893766495001e-08 }, "paired_wilcoxon": { "nonzero_pairs": 1495, "statistic": 425074.0, "two_sided_p_value": 9.774137234129785e-16 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5162520557086544, "bootstrap_se": 0.0018964931708968537, "ci95": [ 0.5125157619285146, 0.5199274005192678 ] }, "negative_mean": { "bootstrap_mean": 0.494874595747303, "bootstrap_se": 0.0022688118570217456, "ci95": [ 0.490479973766417, 0.4993867446995499 ] }, "mean_difference": { "bootstrap_mean": 0.021377459961351344, "bootstrap_se": 0.00230861980474062, "ci95": [ 0.01688002874819331, 0.025961696920935126 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5727880936454849, "bootstrap_se": 0.01284500925313508, "ci95": [ 0.5471571906354515, 0.5986622073578596 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } }, "46": { "positive": { "sample_count": 1495, "sample_mean": 0.5168830953413827, "sample_std": 0.064229599291207, "sample_median": 0.5242176651954651, "mean_pm_std": "0.516883 +/- 0.064230" }, "negative": { "sample_count": 1495, "sample_mean": 0.49942159098526306, "sample_std": 0.07490014925572946, "sample_median": 0.5117457509040833, "mean_pm_std": "0.499422 +/- 0.074900" }, "difference": { "sample_count": 1495, "sample_mean": 0.017461504356119546, "sample_std": 0.07585989833014656, "sample_median": 0.010214507579803467, "mean_pm_std": "0.017462 +/- 0.075860" }, "ordering": { "positive_wins": 853, "negative_wins": 642, "ties": 0, "strict_pairwise_accuracy": 0.5705685618729097, "negative_win_rate": 0.4294314381270903, "tie_rate": 0.0, "cohens_dz": 0.23018096175302127, "exact_sign_test": { "non_tied_pairs": 1495, "two_sided_p_value": 5.3302676191958425e-08 }, "paired_wilcoxon": { "nonzero_pairs": 1495, "statistic": 426756.0, "two_sided_p_value": 2.2107036195762318e-15 } }, "bootstrap": { "positive_mean": { "bootstrap_mean": 0.5168952940156587, "bootstrap_se": 0.0016624667862243931, "ci95": [ 0.5135954594532384, 0.5200632905882836 ] }, "negative_mean": { "bootstrap_mean": 0.4994082986787709, "bootstrap_se": 0.0019270657981015663, "ci95": [ 0.4956341304169052, 0.503140899623467 ] }, "mean_difference": { "bootstrap_mean": 0.01748699533688783, "bootstrap_se": 0.0019659111985969504, "ci95": [ 0.013740118495048487, 0.0214806172196881 ] }, "strict_pairwise_accuracy": { "bootstrap_mean": 0.5704563210702341, "bootstrap_se": 0.012702433919884012, "ci95": [ 0.5451505016722408, 0.5953177257525084 ] }, "tie_rate": { "bootstrap_mean": 0.0, "bootstrap_se": 0.0, "ci95": [ 0.0, 0.0 ] } } } }, "strict_pairwise_accuracy_across_seeds": { "sample_count": 5, "sample_mean": 0.5836789297658862, "sample_std": 0.012740675265080017, "sample_median": 0.585953177257525, "mean_pm_std": "0.583679 +/- 0.012741" }, "bootstrap_of_seed_mean_strict_pairwise_accuracy": { "bootstrap_mean": 0.5837889765886288, "bootstrap_se": 0.010188330082559356, "ci95": [ 0.5638795986622074, 0.6034782608695652 ] } } }, "gatv2_seed_variability": { "seeds": [ 42, 43, 44, 45, 46 ], "metric": "strict_pairwise_accuracy", "values": [ 0.5872909698996656, 0.585953177257525, 0.6020066889632107, 0.5725752508361204, 0.5705685618729097 ], "mean_pm_std": "0.583679 +/- 0.012741" }, "paired_incremental_validity": { "gatv2_minus_length-only": { "strict_pairwise_accuracy_difference": 0.0064214046822741455, "bootstrap": { "bootstrap_mean": 0.006188575250836121, "bootstrap_se": 0.012731036370296255, "ci95": [ -0.01939799331103665, 0.030906354515050166 ] } }, "gatv2_minus_format5": { "strict_pairwise_accuracy_difference": 0.013779264214046738, "bootstrap": { "bootstrap_mean": 0.0137921872909699, "bootstrap_se": 0.013982389367344717, "ci95": [ -0.013377926421404673, 0.04093979933110367 ] } } } } } }