List your company

Science Question Answering on Science QA (Task Acc. & Brier Delta)

65Task AccuracyFINCH
27.76837.43447.156.766May 19, 2026May 27, 2026Jun 4, 2026Jun 13, 2026Jun 21, 2026Jun 29, 2026Jul 8, 2026

Evaluation results

MethodLinks
FINCHModel=Qwen3-4B2026.0565—
Opus-4.72026.0763—
DeepSeek-V4-Pro2026.0763—
SFTModel=Qwen3-4B2026.0562.5—
GPT-5.52026.0760—
FLOWModel=Qwen3-4B2026.0558.3—
Kimi-K2.62026.0758—
L2 RegModel=Qwen3-4B2026.0557.5—
WiSE-FTModel=Qwen3-4B2026.0556.7—
SciReasoner2026.0755—
LoRAModel=Qwen3-4B2026.0551.7—
STMModel=Qwen3-4B2026.0549.2—
TALRModel=Qwen3-4B2026.0548.3—
DFTModel=Qwen3-4B2026.0546.7—
BaseModel=Qwen3-4B2026.0529.2—