List your company

Science Question Answering on ScienceQA (out-of-distribution)

89.04AccuracyCalibRL
86.252886.976487.788.4236Feb 22, 2026

Evaluation results

MethodLinks
CalibRL2026.0289.04
GRPO2026.0288.34
DAPO2026.0287.95
LUFFY2026.0287.95
RL-PLUS2026.0287.5
SFT+GRPO2026.0286.36