When training on GPU, the error "Model diverged with loss = NaN" is often caused by a sotmax that's getting a symbol larger than vocab_size

? ?

轉載于:https://www.cnblogs.com/wuxiangli/p/10344259.html

總結

以上是生活随笔為你收集整理的GPU下train 模型出现nan的全部內容，希望文章能夠幫你解決所遇到的問題。

如果覺得生活随笔網站內容還不錯，歡迎將生活随笔推薦給好友。