List your company

Code Generation on MBPP (Accuracy, Perplexity)

57.4AccuracyMAGS
54.48855.2445656.756May 20, 2026

Evaluation results

MethodLinks
MAGSBase Model=Llama-3.1-8B-Instruct2026.0557.42.97
UnsteeredBase Model=Llama-3.1-8B-Instruct2026.0556.22.978
CDBase Model=Llama-3.1-8B-Instruct2026.0555.52.291
ITIBase Model=Llama-3.1-8B-Instruct2026.0554.82.25
ASBase Model=Llama-3.1-8B-Instruct2026.0554.62.214