List your company

Chat Performance on Alpaca-Eval

55.8ScoreGR3
33.02438.93744.8550.763Mar 11, 2026

Evaluation results

MethodLinks
GR3Model Backbone=Qwen3–8B2026.0355.8765
GRPOModel Backbone=Qwen3–8B2026.0353.51,670
InitialModel Backbone=Qwen3–8B2026.0350.2778
GR3Model Backbone=Qwen3–4B2026.0344.1859
InitialModel Backbone=Qwen3–4B2026.0340.1737
GRPOModel Backbone=Qwen3–4B2026.0333.91,993