Mathematical Reasoning on Theorem
23.8AccuracyCRPS
Evaluation Results
| Method | Links | |
|---|---|---|
| CRPSBase Model=LLaMA3-8B, #Samples=60K2026.04 | 23.8 | |
| CRPSBase Model=LLaMA3-8B, #Samples=15K2026.04 | 23.5 | |
| DARTBase Model=LLaMA3-8B, #Samples=60K2026.04 | 22.9 | |
| CRPSBase Model=LLaMA3-8B, #Samples=30K2026.04 | 21.6 | |
| SIGMABase Model=LLaMA3-8B, #Samples=60K2026.04 | 21.3 | |
| CRPSBase Model=Mistral-7B-v0.1, #Samples=60K2026.04 | 21 | |
| MathFusionBase Model=LLaMA3-8B, #Samples=60K2026.04 | 20 | |
| CRPSBase Model=Mistral-7B-v0.1, #Samples=30K2026.04 | 19.7 | |
| DARTBase Model=LLaMA3-8B, #Samples=590K2026.04 | 19.4 | |
| SIGMABase Model=Mistral-7B-v0.1, #Samples=60K2026.04 | 19.2 | |
| SIGMABase Model=LLaMA3-8B, #Samples=30K2026.04 | 19.1 | |
| DART-Math# Samples=60K, Backbone=Mistral-7B2025.03 | 18.2 | |
| DARTBase Model=Mistral-7B-v0.1, #Samples=60K2026.04 | 18.2 | |
| MathFusion# Samples=60K, Backbone=Mistral-7B2025.03 | 18.1 | |
| MathFusionBase Model=Mistral-7B-v0.1, #Samples=60K2026.04 | 18.1 | |
| SIGMABase Model=Mistral-7B-v0.1, #Samples=30K2026.04 | 18 | |
| CRPSBase Model=Mistral-7B-v0.1, #Samples=15K2026.04 | 17.8 | |
| MathFusion (Sequential)Base Model=LLaMA3-8B, #Samples=30K2026.04 | 17 | |
| DARTBase Model=Mistral-7B-v0.1, #Samples=590K2026.04 | 17 | |
| WizardMathBase Model=Mistral-7B-v0.1, #Samples=418K2026.04 | 16.6 | |
| MMIQCBase Model=LLaMA3-8B, #Samples=2.3M2026.04 | 16.2 | |
| RFTBase Model=Mistral-7B-v0.1, #Samples=590K2026.04 | 16.2 | |
| MathFusion (Sequential)# Samples=30K, Backbone=Mistral-7B2025.03 | 15.5 | |
| MathFusion (Sequential)Base Model=Mistral-7B-v0.1, #Samples=30K2026.04 | 15.5 | |
| MathFusion (Parallel)# Samples=30K, Backbone=Mistral-7B2025.03 | 15.2 | |
| RFTBase Model=LLaMA3-8B, #Samples=590K2026.04 | 14.9 | |
| MetaMathBase Model=Mistral-7B-v0.1, #Samples=400K2026.04 | 14 | |
| MathFusion (DeepSeekMath-7B-RL)# Samples=60K, Backbone=DeepSeekMath-7B-RL2025.03 | 13.8 | |
| MetaMathBase Model=LLaMA3-8B, #Samples=400K2026.04 | 13.8 | |
| MathFusion (Conditional)# Samples=30K, Backbone=Mistral-7B2025.03 | 12.8 | |
| MetaMathBase Model=Mistral-7B-v0.1, #Samples=60K2026.04 | 12.2 | |
| Mosaic-IT# Samples=15K, Backbone=Mistral-7B2025.03 | 9.9 | |
| Mosaic-IT + Original GSM8K and MATH# Samples=30K, Backbone=Mistral-7B2025.03 | 9.5 | |
| GPT Rewritten# Samples=15K, Backbone=Mistral-7B2025.03 | 8.9 | |
| Standard# Samples=15K, Backbone=Mistral-7B2025.03 | 7.6 |