List your company

Math Reasoning on AMC (Accuracy (%))

68.2Accuracy (%)Prefix-RFT
29.82439.78749.7559.713Jul 2, 2025Aug 22, 2025Oct 13, 2025Dec 4, 2025Jan 25, 2026Mar 18, 2026May 9, 2026

Evaluation results

MethodLinks
Prefix-RFTBase Model=Qwen2.5-Math-7B2025.0768.2
LUFFYBase Model=Qwen2.5-Math-7B2025.0765.6
ReLIFTBase Model=Qwen2.5-Math-7B2025.0764.9
GRPOModel=Qwen3-4B-Base, QB (Query-preserved mini batching)=true, RB (Reward-balanced batching)=false2026.0564.4
GRPOModel=Qwen3-4B-Base, QB (Query-preserved mini batching)=true, RB (Reward-balanced batching)=true2026.0564
GRPOModel=Qwen2.5-Math-7B, QB (Query-preserved mini batching)=true, RB (Reward-balanced batching)=true2026.0563.9
SFT+RFTBase Model=Qwen2.5-Math-7B2025.0762.7
RFTBase Model=Qwen2.5-Math-7B2025.0762
GRPOModel=Qwen3-4B-Base, QB (Query-preserved mini batching)=false, RB (Reward-balanced batching)=true2026.0561.5
GRPOModel=Qwen2.5-Math-7B, QB (Query-preserved mini batching)=true, RB (Reward-balanced batching)=false2026.0561.2
Oat-ZeroBase Model=Qwen2.5-Math-7B2025.0761.2
GRPOModel=Qwen2.5-Math-7B, QB (Query-preserved mini batching)=false, RB (Reward-balanced batching)=true2026.0560.1
UFTBase Model=Qwen2.5-Math-7B2025.0758.8
GRPOModel=Qwen3-4B-Base, QB (Query-preserved mini batching)=false, RB (Reward-balanced batching)=false2026.0558
GRPOModel=Qwen2.5-Math-7B, QB (Query-preserved mini batching)=false, RB (Reward-balanced batching)=false2026.0554.9
SimpleRL-ZeroBase Model=Qwen2.5-Math-7B2025.0754.9
SFTBase Model=Qwen2.5-Math-7B2025.0752.8
RL w/ SFT LossBase Model=Qwen2.5-Math-7B2025.0749.7
GRPOModel=Qwen3-1.7B-Base, QB (Query-preserved mini batching)=true, RB (Reward-balanced batching)=true2026.0543.5
GRPOModel=Qwen3-1.7B-Base, QB (Query-preserved mini batching)=false, RB (Reward-balanced batching)=true2026.0542.6
GRPOModel=Qwen3-1.7B-Base, QB (Query-preserved mini batching)=true, RB (Reward-balanced batching)=false2026.0540.2
GRPOModel=Qwen3-1.7B-Base, QB (Query-preserved mini batching)=false, RB (Reward-balanced batching)=false2026.0537
Qwen2.5-Math-7B2025.0731.3