List your company
ResearchFormerly WizWand

Code Generation on HumanEval+, MBPP+, and BigCodeBench Aggregate

70.72Average ScoreCode-A1
50.429655.697360.96566.2327Mar 16, 2026
Updated 7mo ago

Evaluation results

MethodLinks
Code-A1Code LLM=Qwen2.5-Coder-7B-Instruct2026.0370.72
Self-PlayCode LLM=Qwen2.5-Coder-7B-Instruct2026.0370.39
Golden TestsCode LLM=Qwen2.5-Coder-7B-Instruct2026.0370.37
/Code LLM=Qwen2.5-Coder-7B-Instruct2026.0368.35
Code-A1Code LLM=Qwen2.5-Coder-3B-Instruct2026.0366.15
Golden TestsCode LLM=Qwen2.5-Coder-3B-Instruct2026.0365.14
Self-PlayCode LLM=Qwen2.5-Coder-3B-Instruct2026.0364.67
/Code LLM=Qwen2.5-Coder-3B-Instruct2026.0360.84
Code-A1Code LLM=Qwen2.5-Coder-1.5B-Instruct2026.0356.95
Golden TestsCode LLM=Qwen2.5-Coder-1.5B-Instruct2026.0356.23
Self-PlayCode LLM=Qwen2.5-Coder-1.5B-Instruct2026.0355.88
/Code LLM=Qwen2.5-Coder-1.5B-Instruct2026.0351.21