search

mst272 / LLM-Dojo

欢迎来到 LLM-Dojo，这里是一个开源大模型学习场所，使用简洁且易阅读的代码构建模型训练框架(支持各种主流模型如Qwen、Llama、GLM等等)、RLHF框架(DPO/CPO/KTO/PPO)等各种功能。👩‍🎓👨‍🎓
342 stars 30 forks source link

for Qwen-MoE #8

Open BING-LLL opened 3 months ago

BING-LLL commented 3 months ago

I use this tool for Qwen-MoE DPO, but it stopped training at:

return inner_training_loop( args=args, resume_from_checkpoint=resume_from_checkpoint, trial=trial, ignore_keys_for_eval=ignore_keys_for_eval, ) with no output.

mst272 commented 3 months ago

I am running normally on my local machine. but need to update the transformers to the latest version. Please provide more information.