List your company

Code Generation on MBPP (% Avg@4)

77.5Avg@4 (%)GRPO + RePro
57.42862.63967.8573.061Dec 1, 2025

Evaluation results

MethodLinks
GRPO + ReProBackbone=Hunyuan-1.8B-Instruct2025.1277.5
RF++ B + ReProBackbone=Hunyuan-1.8B-Instruct2025.1277
RF++ BBackbone=Hunyuan-1.8B-Instruct2025.1276.3
PPO + ReProBackbone=Hunyuan-1.8B-Instruct2025.1276.2
GRPOBackbone=Hunyuan-1.8B-Instruct2025.1275.9
PPOBackbone=Hunyuan-1.8B-Instruct2025.1275.8
OriginalBackbone=Hunyuan-1.8B-Instruct2025.1273.1
RF++ B + ReProBackbone=Qwen3-1.7B2025.1271.7
PPO + ReProBackbone=Qwen3-1.7B2025.1269.1
GRPO + ReProBackbone=Qwen3-1.7B2025.1268.8
PPOBackbone=Qwen3-1.7B2025.1268.2
GRPOBackbone=Qwen3-1.7B2025.1267.5
OriginalBackbone=Qwen3-1.7B2025.1266.9
RF++ BBackbone=Qwen3-1.7B2025.1266.1
PPO + ReProBackbone=MobileLLM-R1-950M2025.1265.1
PPOBackbone=MobileLLM-R1-950M2025.1263.6
OriginalBackbone=MobileLLM-R1-950M2025.1258.2