使用run_webui，--compile 第一次生成语音会出现报错：RuntimeError: Error: accessing tensor output of CUDAGraphs that has been overwritten by a subsequent run. #894
Unanswered
Morrison-Zhao
asked this question in
Q&A
Replies: 1 comment
-
|
+1 me too when doing --compile on fish speech! fish-speech\\fish_speech\\models\\text2semantic\\inference.py", line 307, in decode_one_token_ar\n codebooks = torch.stack(codebooks, dim=0). To prevent overwriting, clone the tensor outside of torch.compile() or call torch.compiler.cudagraph_mark_step_begin() before each model invocation. |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
我希望自己封装接口实现低延时首帧响应tts，但是测试过程中发现使用compile参数会导致RuntimeError: Error: accessing tensor output of CUDAGraphs that has been overwritten by a subsequent run. Stack trace: File "/data/workspace/morrison/test/ttsStream/fish-speech/fish_speech/models/text2semantic/inference.py", line 305, in decode_one_token_ar
codebooks = torch.stack(codebooks, dim=0). To prevent overwriting, clone the tensor outside of torch.compile() or call torch.compiler.cudagraph_mark_step_begin() before each model invocation.报错，不知道这个问题有无大佬遇到过。希望各位大佬看看有没有解决方案。
Beta Was this translation helpful? Give feedback.
All reactions