Thanks for the good work. Recently I evaluated JetMoE with GSM8K and Humaneval, and I found that the scores I got were higher than you reported. So I want to tell you that the GSM8K I got is 39.4, and the Humaneval I got is 25.6. Maybe this will help.
The scripts I used for evaluation are from https://github.com/QwenLM/Qwen/tree/main/eval.
Thanks for the good work. Recently I evaluated JetMoE with GSM8K and Humaneval, and I found that the scores I got were higher than you reported. So I want to tell you that the GSM8K I got is 39.4, and the Humaneval I got is 25.6. Maybe this will help.
The scripts I used for evaluation are from https://github.com/QwenLM/Qwen/tree/main/eval.