DogeFlow

Results 18 comments of DogeFlow

实时推理?

> 可以，尽量使用多块显卡，而且需要把预加载相关的代码从pipeline中摘出来 已经将preprocess_model、audio_to_coeff、animate_from_coeff这三个模型从pipeline中摘出来了，进行预加载，但还是很慢。如5秒的音频，生成对应的视频需要将近20秒

finetune时的log信息在find CDial-GPT_LCCC-large里面的几个json文件，但download下来的CDial-GPT_LCCC-large里面并没有这几个文件，最后报错，报错信息如下，请问是哪里出了问题

解决方案https://github.com/thu-coai/CDial-GPT/issues/74#issuecomment-912299243

Vall-E X

follow

openvpn connection stops working after upgraded to 2.0.9 with mirrored networking

stay tuned

报错FatalError: `Process abort signal` is detected by the operating system.

同样的错误

FAKE:100

Had the same problem, did you finally solve it

remote-wsl.explorer.addDistro 错误

same issue

Lip shaking problems and partial solutions for wav2lip generated videos 【wav2lip生成的视频唇语抖动问题和部分解决方案】

@fz5400 大佬，抖动的问题解决了吗