List your company

Mathematical Reasoning on DM

51.3AccuracyCRPS
7.51618.88330.2541.617Mar 20, 2025May 23, 2025Jul 27, 2025Sep 30, 2025Dec 4, 2025Feb 7, 2026Apr 13, 2026

Evaluation results

MethodLinks
CRPSBase Model=LLaMA3-8B, #Samples=60K2026.0451.3
CRPSBase Model=LLaMA3-8B, #Samples=30K2026.0449.2
SIGMABase Model=LLaMA3-8B, #Samples=60K2026.0449.2
CRPSBase Model=Mistral-7B-v0.1, #Samples=60K2026.0448.9
DARTBase Model=LLaMA3-8B, #Samples=590K2026.0448
SIGMABase Model=LLaMA3-8B, #Samples=30K2026.0447.5
CRPSBase Model=Mistral-7B-v0.1, #Samples=30K2026.0446.8
SIGMABase Model=Mistral-7B-v0.1, #Samples=60K2026.0446.1
DARTBase Model=Mistral-7B-v0.1, #Samples=590K2026.0445.1
CRPSBase Model=LLaMA3-8B, #Samples=15K2026.0444.5
SIGMABase Model=Mistral-7B-v0.1, #Samples=30K2026.0443.8
MathFusionBase Model=LLaMA3-8B, #Samples=60K2026.0443.4
MathFusion (Sequential)Base Model=LLaMA3-8B, #Samples=30K2026.0442
RFTBase Model=LLaMA3-8B, #Samples=590K2026.0441.7
CRPSBase Model=Mistral-7B-v0.1, #Samples=15K2026.0441.2
MMIQCBase Model=LLaMA3-8B, #Samples=2.3M2026.0441
DARTBase Model=LLaMA3-8B, #Samples=60K2026.0439.9
MathFusion# Samples=60K, Backbone=Mistral-7B2025.0339.2
MathFusionBase Model=Mistral-7B-v0.1, #Samples=60K2026.0439.2
WizardMathBase Model=Mistral-7B-v0.1, #Samples=418K2026.0438.4
MathFusion (DeepSeekMath-7B-RL)# Samples=60K, Backbone=DeepSeekMath-7B-RL2025.0336.5
DART-Math# Samples=60K, Backbone=Mistral-7B2025.0336
DARTBase Model=Mistral-7B-v0.1, #Samples=60K2026.0436
RFTBase Model=Mistral-7B-v0.1, #Samples=590K2026.0435.6
MetaMathBase Model=LLaMA3-8B, #Samples=400K2026.0435
MathFusion (Sequential)# Samples=30K, Backbone=Mistral-7B2025.0329.3
MathFusion (Sequential)Base Model=Mistral-7B-v0.1, #Samples=30K2026.0429.3
MetaMathBase Model=Mistral-7B-v0.1, #Samples=400K2026.0428
MetaMathBase Model=Mistral-7B-v0.1, #Samples=60K2026.0427.2
MathFusion (Parallel)# Samples=30K, Backbone=Mistral-7B2025.0326.5
MathFusion (Conditional)# Samples=30K, Backbone=Mistral-7B2025.0321.4
Standard# Samples=15K, Backbone=Mistral-7B2025.0317
GPT Rewritten# Samples=15K, Backbone=Mistral-7B2025.0313.9
Mosaic-IT + Original GSM8K and MATH# Samples=30K, Backbone=Mistral-7B2025.039.8
Mosaic-IT# Samples=15K, Backbone=Mistral-7B2025.039.2