1{2 "validation_E_single_llm_all_anthropic_mid": {3 "arm": "E_single_llm",4 "committee": "all_anthropic",5 "n": 40,6 "scales": {7 "scale_90": 1.4900272103867178,8 "scale_50": 1.2611647423036356,9 "n_valid": 40,10 "level_90": 0.9,11 "level_50": 0.5,12 "achieved_valid_90": 0.925,13 "achieved_valid_50": 0.525,14 "note": "split conformal, asymmetric scaled residual score, centre = log q50"15 }16 },17 "validation_E_single_llm_all_openai_mid": {18 "arm": "E_single_llm",19 "committee": "all_openai",20 "n": 40,21 "scales": {22 "scale_90": 1.8376879008555251,23 "scale_50": 1.471356381563702,24 "n_valid": 40,25 "level_90": 0.9,26 "level_50": 0.5,27 "achieved_valid_90": 0.925,28 "achieved_valid_50": 0.525,29 "note": "split conformal, asymmetric scaled residual score, centre = log q50"30 }31 },32 "validation_E_single_llm_mixed_random_mid": {33 "arm": "E_single_llm",34 "committee": "mixed_random",35 "n": 40,36 "scales": {37 "scale_90": 1.6294725791155371,38 "scale_50": 1.2643313611844822,39 "n_valid": 40,40 "level_90": 0.9,41 "level_50": 0.5,42 "achieved_valid_90": 0.925,43 "achieved_valid_50": 0.525,44 "note": "split conformal, asymmetric scaled residual score, centre = log q50"45 }46 },47 "validation_F_independent_all_anthropic_mid": {48 "arm": "F_independent",49 "committee": "all_anthropic",50 "n": 40,51 "scales": {52 "scale_90": 1.3966565212756414,53 "scale_50": 1.2851483026993815,54 "n_valid": 40,55 "level_90": 0.9,56 "level_50": 0.5,57 "achieved_valid_90": 0.925,58 "achieved_valid_50": 0.525,59 "note": "split conformal, asymmetric scaled residual score, centre = log q50"60 }61 },62 "validation_F_independent_all_openai_mid": {63 "arm": "F_independent",64 "committee": "all_openai",65 "n": 40,66 "scales": {67 "scale_90": 1.7701450122812816,68 "scale_50": 1.346442789102687,69 "n_valid": 40,70 "level_90": 0.9,71 "level_50": 0.5,72 "achieved_valid_90": 0.925,73 "achieved_valid_50": 0.525,74 "note": "split conformal, asymmetric scaled residual score, centre = log q50"75 }76 },77 "validation_F_independent_mixed_random_mid": {78 "arm": "F_independent",79 "committee": "mixed_random",80 "n": 40,81 "scales": {82 "scale_90": 1.584302968391311,83 "scale_50": 1.2611647423036356,84 "n_valid": 40,85 "level_90": 0.9,86 "level_50": 0.5,87 "achieved_valid_90": 0.925,88 "achieved_valid_50": 0.525,89 "note": "split conformal, asymmetric scaled residual score, centre = log q50"90 }91 },92 "validation_G_debate_r1_mixed_random_mid": {93 "arm": "G_debate_r1",94 "committee": "mixed_random",95 "n": 40,96 "scales": {97 "scale_90": 1.4953562448191138,98 "scale_50": 1.2613728762745116,99 "n_valid": 40,100 "level_90": 0.9,101 "level_50": 0.5,102 "achieved_valid_90": 0.925,103 "achieved_valid_50": 0.525,104 "note": "split conformal, asymmetric scaled residual score, centre = log q50"105 }106 },107 "validation_G_debate_r2_all_anthropic_mid": {108 "arm": "G_debate_r2",109 "committee": "all_anthropic",110 "n": 40,111 "scales": {112 "scale_90": 1.497771158628971,113 "scale_50": 1.3054434024714736,114 "n_valid": 40,115 "level_90": 0.9,116 "level_50": 0.5,117 "achieved_valid_90": 0.925,118 "achieved_valid_50": 0.525,119 "note": "split conformal, asymmetric scaled residual score, centre = log q50"120 }121 },122 "validation_G_debate_r2_all_openai_mid": {123 "arm": "G_debate_r2",124 "committee": "all_openai",125 "n": 40,126 "scales": {127 "scale_90": 1.6689680709853856,128 "scale_50": 1.3139685740963816,129 "n_valid": 40,130 "level_90": 0.9,131 "level_50": 0.5,132 "achieved_valid_90": 0.925,133 "achieved_valid_50": 0.525,134 "note": "split conformal, asymmetric scaled residual score, centre = log q50"135 }136 },137 "validation_G_debate_r2_mixed_random_mid": {138 "arm": "G_debate_r2",139 "committee": "mixed_random",140 "n": 40,141 "scales": {142 "scale_90": 1.5481787277678398,143 "scale_50": 1.2644742287419792,144 "n_valid": 40,145 "level_90": 0.9,146 "level_50": 0.5,147 "achieved_valid_90": 0.925,148 "achieved_valid_50": 0.525,149 "note": "split conformal, asymmetric scaled residual score, centre = log q50"150 }151 },152 "validation_G_debate_r4_mixed_random_mid": {153 "arm": "G_debate_r4",154 "committee": "mixed_random",155 "n": 20,156 "scales": {157 "scale_90": 1.409406786569687,158 "scale_50": 1.0957996076104384,159 "n_valid": 20,160 "level_90": 0.9,161 "level_50": 0.5,162 "achieved_valid_90": 0.95,163 "achieved_valid_50": 0.55,164 "note": "split conformal, asymmetric scaled residual score, centre = log q50"165 }166 },167 "validation_J_cost_matched_mixed_random_mid": {168 "arm": "J_cost_matched",169 "committee": "mixed_random",170 "n": 40,171 "scales": {172 "scale_90": 1.5323539200936531,173 "scale_50": 1.2432409161671234,174 "n_valid": 40,175 "level_90": 0.9,176 "level_50": 0.5,177 "achieved_valid_90": 0.925,178 "achieved_valid_50": 0.525,179 "note": "split conformal, asymmetric scaled residual score, centre = log q50"180 }181 }182}