List your company

Mathematical Reasoning on MATH 500 Hard

80.8AccuracyReAct
48.5656.9365.373.67Dec 4, 2025

Evaluation results

MethodLinks
ReActParadigm=Prompting, Base Model=GPT52025.1280.8
PPOParadigm=Fine-tuning, Base Model=QwQ-32B2025.1273.4
NLACParadigm=Fine-tuning, Base Model=QwQ-32B2025.1272.7
GRPOParadigm=Fine-tuning, Base Model=QwQ-32B2025.1271.8
NLRLParadigm=Fine-tuning, Base Model=QwQ-32B2025.1271.5
NLQLParadigm=Fine-tuning, Base Model=QwQ-32B, Notes=ablation2025.1271.2
Self-DistillationParadigm=Fine-tuning, Base Model=QwQ-32B, Notes=ablation2025.1270.1
RFTParadigm=Fine-tuning, Base Model=QwQ-32B2025.1269.4
Self-DistillationParadigm=Fine-tuning, Base Model=Qwen2.5-7B-Instruct, Notes=ablation2025.1257.1
NLQLParadigm=Fine-tuning, Base Model=Qwen2.5-7B-Instruct, Notes=ablation2025.1256.4
NLACParadigm=Fine-tuning, Base Model=Qwen2.5-7B-Instruct2025.1256.2
NLRLParadigm=Fine-tuning, Base Model=Qwen2.5-7B-Instruct2025.1253.4
GRPOParadigm=Fine-tuning, Base Model=Qwen2.5-7B-Instruct2025.1252.5
PPOParadigm=Fine-tuning, Base Model=Qwen2.5-7B-Instruct2025.1252.3
RFTParadigm=Fine-tuning, Base Model=Qwen2.5-7B-Instruct2025.1249.8