{ "experiment": "R-MATH-INK-06-BEHAVIOR-EXACT-GATE-001", "generated_at": "2026-07-23T20:53:17.837992+00:00", "split_contract": "CROHME trainData writer-validation threshold; testDataGT one-shot", "scope": "truth symbol grouping conditional; c/C, x/X/times, z/Z only", "seeds": [ { "adapter": "research\\runs\\math_ink_06_online_casecontext_refined_seed17_20260723\\skeleton_adapter.pt", "behavior_head": "research\\runs\\math_ink_06_behavior_role_seed17_20260724\\behavior_role_head.pt", "selected_threshold": 0.54, "validation_selected": { "threshold": 0.54, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 340, "abstained": 127, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, "validation_sweep": [ { "threshold": 0.0, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.02, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.04, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.06, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.08, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.1, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.12, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.14, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.16, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.18, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.2, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.22, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.24, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.26, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.28, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.3, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.32, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.34, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.36, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.38, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.4, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.42, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.44, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.46, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.48, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 341, "abstained": 126, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.5, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 340, "abstained": 127, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.52, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 340, "abstained": 127, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.54, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.715203426124197, "gain_pp": 35.76017130620985, "rewrites": 340, "abstained": 127, "beneficial": 170, "harmful": 3, "rewrite_precision": 0.9826589595375722 }, { "threshold": 0.56, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7130620985010707, "gain_pp": 35.54603854389722, "rewrites": 339, "abstained": 128, "beneficial": 169, "harmful": 3, "rewrite_precision": 0.9825581395348837 }, { "threshold": 0.58, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7130620985010707, "gain_pp": 35.54603854389722, "rewrites": 339, "abstained": 128, "beneficial": 169, "harmful": 3, "rewrite_precision": 0.9825581395348837 }, { "threshold": 0.6, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7130620985010707, "gain_pp": 35.54603854389722, "rewrites": 338, "abstained": 129, "beneficial": 169, "harmful": 3, "rewrite_precision": 0.9825581395348837 }, { "threshold": 0.62, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7130620985010707, "gain_pp": 35.54603854389722, "rewrites": 338, "abstained": 129, "beneficial": 169, "harmful": 3, "rewrite_precision": 0.9825581395348837 }, { "threshold": 0.64, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7130620985010707, "gain_pp": 35.54603854389722, "rewrites": 338, "abstained": 129, "beneficial": 169, "harmful": 3, "rewrite_precision": 0.9825581395348837 }, { "threshold": 0.66, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7130620985010707, "gain_pp": 35.54603854389722, "rewrites": 338, "abstained": 129, "beneficial": 169, "harmful": 3, "rewrite_precision": 0.9825581395348837 }, { "threshold": 0.68, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7109207708779444, "gain_pp": 35.33190578158458, "rewrites": 337, "abstained": 130, "beneficial": 168, "harmful": 3, "rewrite_precision": 0.9824561403508771 }, { "threshold": 0.7, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7066381156316917, "gain_pp": 34.903640256959314, "rewrites": 335, "abstained": 132, "beneficial": 166, "harmful": 3, "rewrite_precision": 0.9822485207100592 }, { "threshold": 0.72, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7066381156316917, "gain_pp": 34.903640256959314, "rewrites": 335, "abstained": 132, "beneficial": 166, "harmful": 3, "rewrite_precision": 0.9822485207100592 }, { "threshold": 0.74, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7066381156316917, "gain_pp": 34.903640256959314, "rewrites": 335, "abstained": 132, "beneficial": 166, "harmful": 3, "rewrite_precision": 0.9822485207100592 }, { "threshold": 0.76, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7066381156316917, "gain_pp": 34.903640256959314, "rewrites": 335, "abstained": 132, "beneficial": 166, "harmful": 3, "rewrite_precision": 0.9822485207100592 }, { "threshold": 0.78, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7066381156316917, "gain_pp": 34.903640256959314, "rewrites": 334, "abstained": 133, "beneficial": 166, "harmful": 3, "rewrite_precision": 0.9822485207100592 }, { "threshold": 0.8, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7066381156316917, "gain_pp": 34.903640256959314, "rewrites": 334, "abstained": 133, "beneficial": 166, "harmful": 3, "rewrite_precision": 0.9822485207100592 }, { "threshold": 0.82, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7066381156316917, "gain_pp": 34.903640256959314, "rewrites": 332, "abstained": 135, "beneficial": 165, "harmful": 2, "rewrite_precision": 0.9880239520958084 }, { "threshold": 0.84, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.7066381156316917, "gain_pp": 34.903640256959314, "rewrites": 330, "abstained": 137, "beneficial": 164, "harmful": 1, "rewrite_precision": 0.9939393939393939 }, { "threshold": 0.86, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.6980728051391863, "gain_pp": 34.04710920770878, "rewrites": 326, "abstained": 141, "beneficial": 160, "harmful": 1, "rewrite_precision": 0.9937888198757764 }, { "threshold": 0.88, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.69593147751606, "gain_pp": 33.83297644539615, "rewrites": 322, "abstained": 145, "beneficial": 158, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.9, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.6937901498929336, "gain_pp": 33.61884368308351, "rewrites": 319, "abstained": 148, "beneficial": 157, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.92, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.6895074946466809, "gain_pp": 33.190578158458244, "rewrites": 317, "abstained": 150, "beneficial": 155, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.94, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.6873661670235546, "gain_pp": 32.97644539614561, "rewrites": 315, "abstained": 152, "beneficial": 154, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.96, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.6873661670235546, "gain_pp": 32.97644539614561, "rewrites": 313, "abstained": 154, "beneficial": 154, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.98, "samples": 467, "teacher_exact": 0.3576017130620985, "teacher_visual_family": 0.7301927194860813, "final_exact": 0.6638115631691649, "gain_pp": 30.620985010706637, "rewrites": 293, "abstained": 174, "beneficial": 143, "harmful": 0, "rewrite_precision": 1.0 } ], "official_test": { "threshold": 0.54, "samples": 635, "teacher_exact": 0.4519685039370079, "teacher_visual_family": 0.6425196850393701, "final_exact": 0.584251968503937, "gain_pp": 13.228346456692913, "rewrites": 399, "abstained": 236, "beneficial": 108, "harmful": 24, "rewrite_precision": 0.8181818181818182 } }, { "adapter": "research\\runs\\math_ink_06_online_casecontext_refined_seed31_20260723\\skeleton_adapter.pt", "behavior_head": "research\\runs\\math_ink_06_behavior_role_seed31_20260724\\behavior_role_head.pt", "selected_threshold": 0.9, "validation_selected": { "threshold": 0.9, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5438972162740899, "gain_pp": 35.33190578158458, "rewrites": 250, "abstained": 217, "beneficial": 165, "harmful": 0, "rewrite_precision": 1.0 }, "validation_sweep": [ { "threshold": 0.0, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.02, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.04, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.06, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.08, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.1, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.12, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.14, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.16, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.18, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.2, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.22, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.24, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.26, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.28, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.3, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.32, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.34, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.36, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.38, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.4, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.42, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.44, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.46, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.48, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 256, "abstained": 211, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.5, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 255, "abstained": 212, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.52, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 254, "abstained": 213, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.54, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 254, "abstained": 213, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.56, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 254, "abstained": 213, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.58, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 254, "abstained": 213, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.6, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 254, "abstained": 213, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.62, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 254, "abstained": 213, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.64, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 254, "abstained": 213, "beneficial": 166, "harmful": 2, "rewrite_precision": 0.9880952380952381 }, { "threshold": 0.66, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5396145610278372, "gain_pp": 34.903640256959314, "rewrites": 253, "abstained": 214, "beneficial": 165, "harmful": 2, "rewrite_precision": 0.9880239520958084 }, { "threshold": 0.68, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5396145610278372, "gain_pp": 34.903640256959314, "rewrites": 253, "abstained": 214, "beneficial": 165, "harmful": 2, "rewrite_precision": 0.9880239520958084 }, { "threshold": 0.7, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5396145610278372, "gain_pp": 34.903640256959314, "rewrites": 252, "abstained": 215, "beneficial": 165, "harmful": 2, "rewrite_precision": 0.9880239520958084 }, { "threshold": 0.72, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 251, "abstained": 216, "beneficial": 165, "harmful": 1, "rewrite_precision": 0.9939759036144579 }, { "threshold": 0.74, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 251, "abstained": 216, "beneficial": 165, "harmful": 1, "rewrite_precision": 0.9939759036144579 }, { "threshold": 0.76, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 251, "abstained": 216, "beneficial": 165, "harmful": 1, "rewrite_precision": 0.9939759036144579 }, { "threshold": 0.78, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 251, "abstained": 216, "beneficial": 165, "harmful": 1, "rewrite_precision": 0.9939759036144579 }, { "threshold": 0.8, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 251, "abstained": 216, "beneficial": 165, "harmful": 1, "rewrite_precision": 0.9939759036144579 }, { "threshold": 0.82, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5438972162740899, "gain_pp": 35.33190578158458, "rewrites": 250, "abstained": 217, "beneficial": 165, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.84, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5438972162740899, "gain_pp": 35.33190578158458, "rewrites": 250, "abstained": 217, "beneficial": 165, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.86, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5438972162740899, "gain_pp": 35.33190578158458, "rewrites": 250, "abstained": 217, "beneficial": 165, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.88, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5438972162740899, "gain_pp": 35.33190578158458, "rewrites": 250, "abstained": 217, "beneficial": 165, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.9, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5438972162740899, "gain_pp": 35.33190578158458, "rewrites": 250, "abstained": 217, "beneficial": 165, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.92, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5417558886509636, "gain_pp": 35.11777301927195, "rewrites": 249, "abstained": 218, "beneficial": 164, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.94, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5396145610278372, "gain_pp": 34.903640256959314, "rewrites": 246, "abstained": 221, "beneficial": 163, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.96, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5396145610278372, "gain_pp": 34.903640256959314, "rewrites": 246, "abstained": 221, "beneficial": 163, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.98, "samples": 467, "teacher_exact": 0.1905781584582441, "teacher_visual_family": 0.5481798715203426, "final_exact": 0.5289079229122056, "gain_pp": 33.83297644539615, "rewrites": 238, "abstained": 229, "beneficial": 158, "harmful": 0, "rewrite_precision": 1.0 } ], "official_test": { "threshold": 0.9, "samples": 635, "teacher_exact": 0.27086614173228346, "teacher_visual_family": 0.44724409448818897, "final_exact": 0.3937007874015748, "gain_pp": 12.283464566929133, "rewrites": 270, "abstained": 365, "beneficial": 104, "harmful": 26, "rewrite_precision": 0.8 } }, { "adapter": "research\\runs\\math_ink_06_online_casecontext_refined_seed47_20260723\\skeleton_adapter.pt", "behavior_head": "research\\runs\\math_ink_06_behavior_role_seed47_20260724\\behavior_role_head.pt", "selected_threshold": 0.72, "validation_selected": { "threshold": 0.72, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 272, "abstained": 195, "beneficial": 161, "harmful": 1, "rewrite_precision": 0.9938271604938271 }, "validation_sweep": [ { "threshold": 0.0, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.02, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.04, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.06, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.08, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.1, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.12, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.14, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.16, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.18, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.2, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.22, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.24, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.26, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.28, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.3, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.32, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.34, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.36, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.38, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.4, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.42, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.44, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.46, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.48, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 278, "abstained": 189, "beneficial": 162, "harmful": 2, "rewrite_precision": 0.9878048780487805 }, { "threshold": 0.5, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5845824411134903, "gain_pp": 34.04710920770878, "rewrites": 277, "abstained": 190, "beneficial": 161, "harmful": 2, "rewrite_precision": 0.9877300613496932 }, { "threshold": 0.52, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5845824411134903, "gain_pp": 34.04710920770878, "rewrites": 277, "abstained": 190, "beneficial": 161, "harmful": 2, "rewrite_precision": 0.9877300613496932 }, { "threshold": 0.54, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5845824411134903, "gain_pp": 34.04710920770878, "rewrites": 276, "abstained": 191, "beneficial": 161, "harmful": 2, "rewrite_precision": 0.9877300613496932 }, { "threshold": 0.56, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5845824411134903, "gain_pp": 34.04710920770878, "rewrites": 274, "abstained": 193, "beneficial": 161, "harmful": 2, "rewrite_precision": 0.9877300613496932 }, { "threshold": 0.58, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5845824411134903, "gain_pp": 34.04710920770878, "rewrites": 274, "abstained": 193, "beneficial": 161, "harmful": 2, "rewrite_precision": 0.9877300613496932 }, { "threshold": 0.6, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 273, "abstained": 194, "beneficial": 161, "harmful": 1, "rewrite_precision": 0.9938271604938271 }, { "threshold": 0.62, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 273, "abstained": 194, "beneficial": 161, "harmful": 1, "rewrite_precision": 0.9938271604938271 }, { "threshold": 0.64, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 273, "abstained": 194, "beneficial": 161, "harmful": 1, "rewrite_precision": 0.9938271604938271 }, { "threshold": 0.66, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 273, "abstained": 194, "beneficial": 161, "harmful": 1, "rewrite_precision": 0.9938271604938271 }, { "threshold": 0.68, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 272, "abstained": 195, "beneficial": 161, "harmful": 1, "rewrite_precision": 0.9938271604938271 }, { "threshold": 0.7, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 272, "abstained": 195, "beneficial": 161, "harmful": 1, "rewrite_precision": 0.9938271604938271 }, { "threshold": 0.72, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5867237687366167, "gain_pp": 34.261241970021416, "rewrites": 272, "abstained": 195, "beneficial": 161, "harmful": 1, "rewrite_precision": 0.9938271604938271 }, { "threshold": 0.74, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5845824411134903, "gain_pp": 34.04710920770878, "rewrites": 271, "abstained": 196, "beneficial": 160, "harmful": 1, "rewrite_precision": 0.9937888198757764 }, { "threshold": 0.76, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5845824411134903, "gain_pp": 34.04710920770878, "rewrites": 270, "abstained": 197, "beneficial": 160, "harmful": 1, "rewrite_precision": 0.9937888198757764 }, { "threshold": 0.78, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.582441113490364, "gain_pp": 33.83297644539615, "rewrites": 269, "abstained": 198, "beneficial": 159, "harmful": 1, "rewrite_precision": 0.99375 }, { "threshold": 0.8, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5802997858672377, "gain_pp": 33.61884368308351, "rewrites": 268, "abstained": 199, "beneficial": 158, "harmful": 1, "rewrite_precision": 0.9937106918238994 }, { "threshold": 0.82, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5802997858672377, "gain_pp": 33.61884368308351, "rewrites": 268, "abstained": 199, "beneficial": 158, "harmful": 1, "rewrite_precision": 0.9937106918238994 }, { "threshold": 0.84, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5802997858672377, "gain_pp": 33.61884368308351, "rewrites": 268, "abstained": 199, "beneficial": 158, "harmful": 1, "rewrite_precision": 0.9937106918238994 }, { "threshold": 0.86, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5802997858672377, "gain_pp": 33.61884368308351, "rewrites": 267, "abstained": 200, "beneficial": 158, "harmful": 1, "rewrite_precision": 0.9937106918238994 }, { "threshold": 0.88, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.576017130620985, "gain_pp": 33.190578158458244, "rewrites": 263, "abstained": 204, "beneficial": 155, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.9, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5717344753747323, "gain_pp": 32.76231263383298, "rewrites": 261, "abstained": 206, "beneficial": 153, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.92, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5653104925053534, "gain_pp": 32.11991434689507, "rewrites": 256, "abstained": 211, "beneficial": 150, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.94, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5610278372591007, "gain_pp": 31.69164882226981, "rewrites": 253, "abstained": 214, "beneficial": 148, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.96, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5546038543897216, "gain_pp": 31.049250535331907, "rewrites": 250, "abstained": 217, "beneficial": 145, "harmful": 0, "rewrite_precision": 1.0 }, { "threshold": 0.98, "samples": 467, "teacher_exact": 0.24411134903640258, "teacher_visual_family": 0.5952890792291221, "final_exact": 0.5417558886509636, "gain_pp": 29.764453961456102, "rewrites": 238, "abstained": 229, "beneficial": 139, "harmful": 0, "rewrite_precision": 1.0 } ], "official_test": { "threshold": 0.72, "samples": 635, "teacher_exact": 0.2866141732283465, "teacher_visual_family": 0.5007874015748032, "final_exact": 0.4409448818897638, "gain_pp": 15.433070866141732, "rewrites": 307, "abstained": 328, "beneficial": 125, "harmful": 27, "rewrite_precision": 0.8223684210526315 } } ], "official_test_summary": { "teacher_exact": { "values": [ 0.4519685039370079, 0.27086614173228346, 0.2866141732283465 ], "mean": 0.33648293963254594 }, "teacher_visual_family": { "values": [ 0.6425196850393701, 0.44724409448818897, 0.5007874015748032 ], "mean": 0.5301837270341206 }, "final_exact": { "values": [ 0.584251968503937, 0.3937007874015748, 0.4409448818897638 ], "mean": 0.47296587926509187 }, "gain_pp": { "values": [ 13.228346456692913, 12.283464566929133, 15.433070866141732 ], "mean": 13.648293963254593 }, "rewrite_precision": { "values": [ 0.8181818181818182, 0.8, 0.8223684210526315 ], "mean": 0.8135167464114832 } }, "track": "R_noncommercial_only", "product_validation": false, "distillation_allowed": false }