List your company

Mathematical Reasoning on GSM8K (test) (Pass@1)

86.4Pass@1LEMMA (w/ MetaMath)
64.66470.30775.9581.593Mar 21, 2025

Evaluation results

MethodLinks
LEMMA (w/ MetaMath)Backbone=LLaMA3-8B, # Samples=403.59k2025.0386.4
LEMMA (w/ MetaMath)Backbone=DeepSeekMath-7B, # Samples=403.59k2025.0383
S³C-Math (w/ MetaMath)Backbone=LLaMA3-8B, # Samples=927k2025.0382.9
S³C-Math (w/ MetaMath)Backbone=DeepSeekMath-7B, # Samples=927k2025.0382.5
LEMMABackbone=DeepSeekMath-7B, # Samples=88.90k2025.0380.4
MetaMathBackbone=DeepSeekMath-7B, # Samples=394.99k2025.0379.4
MetaMathBackbone=LLaMA3-8B, # Samples=394.99k2025.0379.2
LEMMABackbone=LLaMA3-8B, # Samples=88.90k2025.0379.2
GPTAugBackbone=DeepSeekMath-7B, # Samples=88.62k2025.0377.8
RefAug-90kBackbone=LLaMA3-8B, # Samples=89.92k2025.0377.4
RefAug-90kBackbone=DeepSeekMath-7B, # Samples=89.92k2025.0376.7
RefAugBackbone=LLaMA3-8B, # Samples=29.94k2025.0375.9
RefAugBackbone=DeepSeekMath-7B, # Samples=29.94k2025.0375.5
RFTBackbone=DeepSeekMath-7B, # Samples=86.52k2025.0373.3
GPTAugBackbone=LLaMA3-8B, # Samples=88.62k2025.0372.1
ISCBackbone=LLaMA3-8B, # Samples=86.78k2025.0370.8
SFTBackbone=DeepSeekMath-7B, # Samples=14.97k2025.0368.1
RFTBackbone=LLaMA3-8B, # Samples=86.52k2025.0367.3
ISCBackbone=DeepSeekMath-7B, # Samples=86.78k2025.0366.3
SFTBackbone=LLaMA3-8B, # Samples=14.97k2025.0365.5