List your company

Mathematical Reasoning on GSM8K (Accuracy, TKN)

94.8AccuracyCLSR
91.47292.33693.294.064Jun 28, 2026

Evaluation results

MethodLinks
CLSRBackbone=Qwen3-8B, T (number of rounds)=32026.0694.8214
PBrdBackbone=Qwen3-8B2026.0693.6235
CLSRBackbone=Qwen3-8B2026.0692.890
PALBackbone=Qwen3-8B2026.0692.5148
P2SBackbone=Qwen3-8B2026.0692.2316
PoTBackbone=Qwen3-8B2026.0692.1113
CLSRBackbone=Qwen3-8B, T (number of rounds)=12026.0692.183
CoTBackbone=Qwen3-8B2026.0691.6248