List your company

Abstention on ABSTAINEQA HM3D-380

71.1F1 ScoreSEMANTIC FLIP
29.385640.215351.04561.8747Jun 15, 2026

Evaluation results

MethodLinks
SEMANTIC FLIPParams=7B + small head2026.0671.166.8481.5852.11
Qwen-32B-CoarseParams=32B2026.0667.4656.8489.4724.21
Qwen-32B-FineParams=32B2026.0667.4153.9595.2612.63
Qwen-32B-Fine + Naive CoTParams=32B2026.0630.9956.5419.693.44