List your company

Math Reasoning on ThmQA

34.3Multi@5 AccuracyDRIFT-5turn
21.50824.82928.1531.471May 29, 2026

Evaluation results

MethodLinks
DRIFT-5turnBackbone=Qwen2.5-3B-Instruct, Protocol=Ours, Max turns=52026.0534.3
UFO-5turnBackbone=Qwen2.5-3B-Instruct, Protocol=Multi-turn baselines (Online), Max turns=52026.0533.5
SFT-5turnBackbone=Qwen2.5-3B-Instruct, Protocol=Multi-turn baselines (Offline), Max turns=52026.0533.1
SCoRe-2turnBackbone=Qwen2.5-3B-Instruct, Protocol=Multi-turn baselines (Online), Max turns=22026.0531.8
UFO-5turnBackbone=Llama3.1-8B-Instruct, Protocol=Multi-turn baselines (Online), Max turns=52026.0530.3
PPOBackbone=Llama3.1-8B-Instruct, Protocol=Single-turn baselines2026.0530.1
DRIFT-5turnBackbone=Llama3.1-8B-Instruct, Protocol=Ours, Max turns=52026.0529.9
SFTBackbone=Qwen2.5-3B-Instruct, Protocol=Single-turn baselines2026.0529.8
STaR-2turnBackbone=Qwen2.5-3B-Instruct, Protocol=Multi-turn baselines (Offline), Max turns=22026.0529.5
PPOBackbone=Qwen2.5-3B-Instruct, Protocol=Single-turn baselines2026.0529
STaR-2turnBackbone=Llama3.1-8B-Instruct, Protocol=Multi-turn baselines (Offline), Max turns=22026.0526.2
SCoRe-2turnBackbone=Llama3.1-8B-Instruct, Protocol=Multi-turn baselines (Online), Max turns=22026.0526.1
BaseBackbone=Qwen2.5-3B-Instruct, Protocol=Single-turn baselines2026.0526
SFTBackbone=Llama3.1-8B-Instruct, Protocol=Single-turn baselines2026.0524.3
SFT-5turnBackbone=Llama3.1-8B-Instruct, Protocol=Multi-turn baselines (Offline), Max turns=52026.0523.3
BaseBackbone=Llama3.1-8B-Instruct, Protocol=Single-turn baselines2026.0522