{"version":"V2.2","date":"2026-10-09","T":4040,"Q":4040,"G":3690,"models":7,"reviewed_paths":5503,"reviewed_unique_tasks":5378,"quality":{"KEEP":4047,"WEAK":1120,"UNRESOLVED":302,"BAD":34},"phases":[{"phase":"V2","reviewed":5378,"quality":{"KEEP":3960,"WEAK":1095,"UNRESOLVED":298,"BAD":25},"exported":3953,"technical_KEEP_excluded":7},{"phase":"V2.1","reviewed":37,"quality":{"KEEP":26,"WEAK":7,"BAD":3,"UNRESOLVED":1},"exported":26,"technical_KEEP_excluded":0},{"phase":"V2.2","reviewed":88,"quality":{"BAD":6,"KEEP":61,"WEAK":18,"UNRESOLVED":3},"exported":61,"technical_KEEP_excluded":0}],"lengths":{"paths":4040,"direct":2132,"one_crop":1146,"two_plus_crops":762,"steps":{"n":4040,"total":7047,"mean":1.744,"median":1.0,"p90":3,"max":9},"processed_tokens":{"n":4040,"total":10770105,"mean":2665.868,"median":2270.0,"p90":3856,"max":8162},"target_tokens":{"n":4040,"total":652944,"mean":161.62,"median":144.0,"p90":315,"max":1001}},"crop_length":{"n":4040,"total":3007,"mean":0.744,"median":0.0,"p90":2,"max":8},"step_histogram":{"1":2132,"2":1146,"3":535,"4":154,"5":49,"6":14,"7":8,"8":1,"9":1},"by_model":{"Luna":{"paths":1113,"direct":839,"one_crop":205,"two_plus_crops":69,"steps":{"n":1113,"total":1469,"mean":1.32,"median":1,"p90":2,"max":5},"processed_tokens":{"n":1113,"total":2596910,"mean":2333.252,"median":2126,"p90":3061,"max":5509},"target_tokens":{"n":1113,"total":116152,"mean":104.359,"median":68,"p90":188,"max":504}},"Sol 5.6":{"paths":1081,"direct":729,"one_crop":227,"two_plus_crops":125,"steps":{"n":1081,"total":1617,"mean":1.496,"median":1,"p90":3,"max":6},"processed_tokens":{"n":1081,"total":2650870,"mean":2452.239,"median":2129,"p90":3325,"max":7121},"target_tokens":{"n":1081,"total":141289,"mean":130.702,"median":91,"p90":270,"max":835}},"GLM":{"paths":514,"direct":58,"one_crop":313,"two_plus_crops":143,"steps":{"n":514,"total":1177,"mean":2.29,"median":2.0,"p90":3,"max":7},"processed_tokens":{"n":514,"total":1605150,"mean":3122.86,"median":3060.0,"p90":4114,"max":6331},"target_tokens":{"n":514,"total":133101,"mean":258.951,"median":219.5,"p90":418,"max":1001}},"Grok":{"paths":587,"direct":30,"one_crop":203,"two_plus_crops":354,"steps":{"n":587,"total":1680,"mean":2.862,"median":3,"p90":4,"max":9},"processed_tokens":{"n":587,"total":2089344,"mean":3559.359,"median":3442,"p90":4776,"max":8162},"target_tokens":{"n":587,"total":150306,"mean":256.058,"median":248,"p90":386,"max":780}},"Astra":{"paths":382,"direct":277,"one_crop":94,"two_plus_crops":11,"steps":{"n":382,"total":503,"mean":1.317,"median":1.0,"p90":2,"max":6},"processed_tokens":{"n":382,"total":880041,"mean":2303.772,"median":2142.5,"p90":2866,"max":5086},"target_tokens":{"n":382,"total":43576,"mean":114.073,"median":98.5,"p90":183,"max":604}},"Gemini":{"paths":248,"direct":131,"one_crop":71,"two_plus_crops":46,"steps":{"n":248,"total":421,"mean":1.698,"median":1.0,"p90":3,"max":6},"processed_tokens":{"n":248,"total":659692,"mean":2660.048,"median":2293.0,"p90":3621,"max":6922},"target_tokens":{"n":248,"total":51802,"mean":208.879,"median":188.5,"p90":366,"max":800}},"K3":{"paths":115,"direct":68,"one_crop":33,"two_plus_crops":14,"steps":{"n":115,"total":180,"mean":1.565,"median":1,"p90":3,"max":4},"processed_tokens":{"n":115,"total":288098,"mean":2505.2,"median":2171,"p90":3250,"max":4615},"target_tokens":{"n":115,"total":16718,"mean":145.374,"median":115,"p90":276,"max":489}}},"by_source":{"mme_realworld":{"paths":2669,"direct":1149,"one_crop":948,"two_plus_crops":572,"steps":{"n":2669,"total":5008,"mean":1.876,"median":2,"p90":3,"max":8},"processed_tokens":{"n":2669,"total":7484721,"mean":2804.317,"median":2648,"p90":4015,"max":8162},"target_tokens":{"n":2669,"total":428839,"mean":160.674,"median":148,"p90":322,"max":1001}},"mmmu_pro":{"paths":549,"direct":512,"one_crop":14,"two_plus_crops":23,"steps":{"n":549,"total":616,"mean":1.122,"median":1,"p90":1,"max":4},"processed_tokens":{"n":549,"total":1194925,"mean":2176.548,"median":2134,"p90":2509,"max":4689},"target_tokens":{"n":549,"total":82632,"mean":150.514,"median":123,"p90":270,"max":835}},"visreason":{"paths":240,"direct":206,"one_crop":9,"two_plus_crops":25,"steps":{"n":240,"total":318,"mean":1.325,"median":1.0,"p90":3,"max":6},"processed_tokens":{"n":240,"total":538760,"mean":2244.833,"median":2123.5,"p90":2942,"max":6239},"target_tokens":{"n":240,"total":41691,"mean":173.713,"median":125.5,"p90":346,"max":1001}},"vstar":{"paths":69,"direct":13,"one_crop":43,"two_plus_crops":13,"steps":{"n":69,"total":141,"mean":2.043,"median":2,"p90":3,"max":5},"processed_tokens":{"n":69,"total":187049,"mean":2710.855,"median":2613,"p90":3231,"max":4203},"target_tokens":{"n":69,"total":11456,"mean":166.029,"median":158,"p90":272,"max":463}},"worldbench":{"paths":513,"direct":252,"one_crop":132,"two_plus_crops":129,"steps":{"n":513,"total":964,"mean":1.879,"median":2,"p90":3,"max":9},"processed_tokens":{"n":513,"total":1364650,"mean":2660.136,"median":2227,"p90":3930,"max":7531},"target_tokens":{"n":513,"total":88326,"mean":172.175,"median":148,"p90":363,"max":777}}},"by_material":{"academic":{"paths":572,"direct":531,"one_crop":14,"two_plus_crops":27,"steps":{"n":572,"total":649,"mean":1.135,"median":1.0,"p90":1,"max":5},"processed_tokens":{"n":572,"total":1248749,"mean":2183.128,"median":2134.0,"p90":2510,"max":4689},"target_tokens":{"n":572,"total":86204,"mean":150.706,"median":122.0,"p90":270,"max":835}},"chart_table":{"paths":1038,"direct":615,"one_crop":252,"two_plus_crops":171,"steps":{"n":1038,"total":1663,"mean":1.602,"median":1.0,"p90":3,"max":5},"processed_tokens":{"n":1038,"total":2778161,"mean":2676.456,"median":2157.0,"p90":4045,"max":6056},"target_tokens":{"n":1038,"total":154346,"mean":148.696,"median":84.0,"p90":307,"max":628}},"count":{"paths":329,"direct":82,"one_crop":86,"two_plus_crops":161,"steps":{"n":329,"total":864,"mean":2.626,"median":2,"p90":4,"max":9},"processed_tokens":{"n":329,"total":1121386,"mean":3408.468,"median":3113,"p90":5037,"max":8162},"target_tokens":{"n":329,"total":86672,"mean":263.441,"median":250,"p90":476,"max":1001}},"map":{"paths":4,"direct":3,"one_crop":1,"two_plus_crops":0,"steps":{"n":4,"total":5,"mean":1.25,"median":1.0,"p90":2,"max":2},"processed_tokens":{"n":4,"total":8724,"mean":2181,"median":2110.0,"p90":2537,"max":2537},"target_tokens":{"n":4,"total":427,"mean":106.75,"median":81.0,"p90":188,"max":188}},"natural":{"paths":959,"direct":365,"one_crop":308,"two_plus_crops":286,"steps":{"n":959,"total":1983,"mean":2.068,"median":2,"p90":4,"max":7},"processed_tokens":{"n":959,"total":2745926,"mean":2863.322,"median":2662,"p90":4051,"max":7869},"target_tokens":{"n":959,"total":165668,"mean":172.751,"median":155,"p90":342,"max":777}},"text":{"paths":898,"direct":330,"one_crop":476,"two_plus_crops":92,"steps":{"n":898,"total":1565,"mean":1.743,"median":2.0,"p90":3,"max":4},"processed_tokens":{"n":898,"total":2328399,"mean":2592.872,"median":2627.0,"p90":3126,"max":4194},"target_tokens":{"n":898,"total":117936,"mean":131.332,"median":145.0,"p90":224,"max":417}},"visual_puzzle":{"paths":240,"direct":206,"one_crop":9,"two_plus_crops":25,"steps":{"n":240,"total":318,"mean":1.325,"median":1.0,"p90":3,"max":6},"processed_tokens":{"n":240,"total":538760,"mean":2244.833,"median":2123.5,"p90":2942,"max":6239},"target_tokens":{"n":240,"total":41691,"mean":173.713,"median":125.5,"p90":346,"max":1001}}},"audit_by_model":{"Luna":{"reviewed":1286,"quality":{"KEEP":1113,"WEAK":128,"BAD":4,"UNRESOLVED":41},"exported":1113},"Sol 5.6":{"reviewed":1279,"quality":{"KEEP":1081,"UNRESOLVED":78,"WEAK":116,"BAD":4},"exported":1081},"GLM":{"reviewed":953,"quality":{"KEEP":521,"WEAK":367,"UNRESOLVED":57,"BAD":8},"exported":514},"Grok":{"reviewed":1094,"quality":{"WEAK":452,"KEEP":587,"UNRESOLVED":47,"BAD":8},"exported":587},"Astra":{"reviewed":443,"quality":{"KEEP":382,"UNRESOLVED":55,"WEAK":5,"BAD":1},"exported":382},"Gemini":{"reviewed":293,"quality":{"KEEP":248,"WEAK":25,"UNRESOLVED":20},"exported":248},"K3":{"reviewed":155,"quality":{"KEEP":115,"WEAK":27,"BAD":9,"UNRESOLVED":4},"exported":115}},"audit_by_source":{"mme_realworld":{"reviewed":3552,"quality":{"KEEP":2676,"WEAK":730,"UNRESOLVED":131,"BAD":15},"exported":2669},"mmmu_pro":{"reviewed":721,"quality":{"WEAK":111,"KEEP":549,"UNRESOLVED":58,"BAD":3},"exported":549},"visreason":{"reviewed":376,"quality":{"WEAK":68,"KEEP":240,"BAD":6,"UNRESOLVED":62},"exported":240},"vstar":{"reviewed":95,"quality":{"WEAK":26,"KEEP":69},"exported":69},"worldbench":{"reviewed":759,"quality":{"WEAK":185,"KEEP":513,"UNRESOLVED":51,"BAD":10},"exported":513}},"behaviors":{"locate_then_detail":1545,"cross_region_compare_or_bind":624,"honest_miss_then_switch":133,"direct_when_sufficient":2132},"crop_gains":{"new_detail":1692,"comparison_evidence":320,"honest_miss":79,"informative_negative":233,"binding_context":683},"historical_advice_paths":3948,"historical_attribution_paths":{"Opus 5.5":380,"Sonnet 5.5":3564,"Sonnet workflow (version from batch label)":4},"human_gold":false,"training_run":false,"all_selected_have_process_KEEP":true,"unreviewed_selected":0,"source_inputs":{"v2_merged_keep_trajectories_20261008.jsonl":"6f359c3dc5e0614f9d7a9c76aff1041994139995f7a48160e17c93f417fb833d","v2_merged_review_table_20261008.jsonl":"b8871f3961e229a469cd444b6fc62ac225221c3fe48a0a5b2a7590de674fd221","v2_1_increment_keep_20261008.jsonl":"25298b162f85837cd0906505efd60b2d1fe9b56d214d0d936a815fcc925aac41","v2_1_topup_review_table_20261008.jsonl":"bcd692736f2f26cd80af747b4ac228bcce546c71d406ebb9f2691ab123e94c94","v2_2_increment_keep_20261009.jsonl":"276d5bcbd4f575bb4e573393724ca8dc1515f2401ba404d81f8ff33719887369","v2_2_topup_review_table_20261009.jsonl":"f1fa848e02ae8630a6cf7349bcfe231493a74dfc4c33fb938f1e72501880b7d8"},"groups_with_multiple_tasks":301}
