Jerry

Results 6 comments of Jerry

RuntimeError: Expected object of backend CUDA but got backend CPU for argument #3 'index'

> I fixed this by changing > fixed_embeddings = encoder(fixed_captions) line > to > fixed_embeddings = encoder(fixed_captions.to(device)) > > But im not sure if this is a good way to...

Use this repo or wait for v2?

Well, maybe we have no T2F v2...

about the center loss

Same confusion with you.

您好，我想知道如何利用训练结果去预测新数据

同问

关于tokenizer.tokenize的疑问

请问解决了吗，我也对齐不了……

loss 训练值过大

我也是！loss都到1w+了，感觉不太正常。怎么解决？