List your company

Mathematical Reasoning on Countdown-34 (held-out)

81.26AccuracyDS (Oracle)
69.414472.489775.56578.6403Mar 11, 2026

Evaluation results

MethodLinks
DS (Oracle)Backbone=Qwen2.5-7B, Rollouts=1006k, Finetuned=true2026.0381.26
DPSBackbone=Qwen2.5-7B, Rollouts=246k, Finetuned=true2026.0381.15
HRBackbone=Qwen2.5-7B, Rollouts=246k, Finetuned=true2026.0378.15
USBackbone=Qwen2.5-7B, Rollouts=246k, Finetuned=true2026.0377.84
DS (Oracle)Backbone=Qwen2.5-3B, Rollouts=1141k, Finetuned=true2026.0374.95
DPSBackbone=Qwen2.5-3B, Rollouts=246k, Finetuned=true2026.0374.27
HRBackbone=Qwen2.5-3B, Rollouts=246k, Finetuned=true2026.0370.19
USBackbone=Qwen2.5-3B, Rollouts=246k, Finetuned=true2026.0369.87