Self Checks

• This template is only for bug reports. For questions, please visit Discussions.
• I have thoroughly reviewed the project documentation (installation, training, inference) but couldn't find information to solve my problem. English 中文 日本語 Portuguese (Brazil)
• I have searched for existing issues, including closed ones. Search issues
• I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
• [FOR CHINESE USERS] 请务必使用英文提交 Issue，否则会被关闭。谢谢！:）
• Please do not modify this template and fill in all required fields.

Cloud or Self Hosted

Self Hosted (Source)

Environment Details

python3.10 torch2.3.0+cuda12.1 fish-speech-1.5

Steps to Reproduce

As it is called, the gpu memory will increase irregularly from time to time and will not be released actively. How can I limit the following processing length? Can I set the maximum number of processing tokens for splitting?

Another issue:

<frozen importlib._bootstrap_external>:883: _call_with_frames_removed: block: [0,0,0], thread: [191,0,0] Assertion `index out of bounds: 0 <= tmp4 < 1024` failed.

✔️ Expected Behavior

Stable GPU utilization

❌ Actual Behavior

Random increase in GPU utilization until the memory is exhausted or the following errors occur:

<frozen importlib._bootstrap_external>:883: _call_with_frames_removed: block: [0,0,0], thread: [191,0,0] Assertion `index out of bounds: 0 <= tmp4 < 1024` failed.