List your company

Reasoning on AMC23 (Acc, Len, Faith)

95AccuracyGeoFaith
1737.2557.577.75May 26, 2026

Evaluation results

MethodLinks
GeoFaithBackbone=Qwen3-4B2026.05952.295.6
GRPOBackbone=Qwen3-4B2026.0592.53.386.9
KnowRLBackbone=Qwen3-4B2026.0592.52.989.6
TruthRLBackbone=Qwen3-4B2026.0592.51.789.2
THSBackbone=Qwen3-4B2026.05902.685.9
OriginalBackbone=Qwen3-4B2026.0582.54.692.5
GeoFaithBackbone=Qwen3-1.7B2026.05451.254.5
OriginalBackbone=Qwen3-1.7B2026.0537.52.348.6
GRPOBackbone=Qwen3-1.7B2026.0537.5151.3
KnowRLBackbone=Qwen3-1.7B2026.0532.5142.6
THSBackbone=Qwen3-1.7B2026.05250.433.3
TruthRLBackbone=Qwen3-1.7B2026.05200.823.6