微调模型时train和eval的loss和梯度有时是nan

FunAudioLLM / CosyVoice

Multi-lingual large voice generation model, providing inference, training and deployment full-stack ability.

https://funaudiollm.github.io/

Apache License 2.0

6.54k stars 703 forks source link

Open ScottishFold007 opened 4 days ago

ScottishFold007 commented 4 days ago

在微调模型的时候，train和eval的loss和梯度有时是nan，这个情况是否正常？可能的原因有哪些呢？

aluminumbox commented 3 days ago

不是一直nan就可以