| "# Stage 3 downstream: 10-task lm-eval, 0-shot, acc+acc_norm+stderr. Cells: base-free / base-impose-R8 / CE-adapted-R8(merged, 0.8147@R8)." |
| "# dense-bracket (orch 0122): OLMo-1B-0724-hf (d7cbab74) + OLMo-7B-0724-hf (1ee306df), free routing, same 10-task 0-shot harness. Era-matched released checkpoints, NOT data-matched (OLMo-1B ~3T Dolma vs OLMoE 5.1T)." |
| task,metric,base_free,base_free_se,impose_R8,impose_R8_se,CE_adapt_R8,CE_adapt_R8_se,CEadapt_minus_base,impose_minus_base,olmo1b_free,olmo1b_se,olmo7b_free,olmo7b_se
|
| arc_easy,acc,0.7715,0.0086,0.2799,0.0092,0.6688,0.0097,-0.1027,-0.4916,0.6077,0.0100,0.7319,0.0091
|
| arc_easy,acc_norm,0.7664,0.0087,0.2706,0.0091,0.6393,0.0099,-0.1271,-0.4958,0.5665,0.0102,0.7441,0.0090
|
| arc_challenge,acc,0.4701,0.0146,0.2150,0.0120,0.3396,0.0138,-0.1305,-0.2551,0.3038,0.0134,0.4044,0.0143
|
| arc_challenge,acc_norm,0.4991,0.0146,0.2526,0.0127,0.3677,0.0141,-0.1314,-0.2466,0.3208,0.0136,0.4360,0.0145
|
| hellaswag,acc,0.5856,0.0049,0.2572,0.0044,0.4606,0.0050,-0.1251,-0.3284,0.4863,0.0050,0.5822,0.0049
|
| hellaswag,acc_norm,0.7822,0.0041,0.2657,0.0044,0.6211,0.0048,-0.1611,-0.5165,0.6606,0.0047,0.7786,0.0041
|
| piqa,acc,0.7873,0.0095,0.5180,0.0117,0.7285,0.0104,-0.0588,-0.2693,0.7486,0.0101,0.7987,0.0094
|
| piqa,acc_norm,0.7976,0.0094,0.4973,0.0117,0.7372,0.0103,-0.0604,-0.3003,0.7508,0.0101,0.8047,0.0092
|
| winogrande,acc,0.6922,0.0130,0.4909,0.0141,0.5706,0.0139,-0.1215,-0.2013,0.6133,0.0137,0.6740,0.0132
|
| openbookqa,acc,0.3220,0.0209,0.1460,0.0158,0.2400,0.0191,-0.0820,-0.1760,0.2420,0.0192,0.3020,0.0206
|
| openbookqa,acc_norm,0.4480,0.0223,0.2660,0.0198,0.3620,0.0215,-0.0860,-0.1820,0.3560,0.0214,0.4120,0.0220
|
| sciq,acc,0.9370,0.0077,0.2930,0.0144,0.9250,0.0083,-0.0120,-0.6440,0.9260,0.0083,0.9630,0.0060
|
| sciq,acc_norm,0.9260,0.0083,0.2700,0.0140,0.8900,0.0099,-0.0360,-0.6560,0.9190,0.0086,0.9560,0.0065
|
| boolq,acc,0.7018,0.0080,0.4037,0.0086,0.6346,0.0084,-0.0673,-0.2982,0.6612,0.0083,0.7749,0.0073
|
| lambada_openai,acc,0.7056,0.0063,0.0000,,0.5703,0.0069,-0.1353,-0.7056,0.6074,0.0068,0.7128,0.0063
|
| copa,acc,0.8500,0.0359,0.5600,0.0499,0.7500,0.0435,-0.1000,-0.2900,0.8100,0.0394,0.8300,0.0378
|
|
|