Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

llm+flow微调 #1067

Open
Sunnycl opened this issue Mar 12, 2025 · 2 comments
Open

llm+flow微调 #1067

Sunnycl opened this issue Mar 12, 2025 · 2 comments

Comments

@Sunnycl
Copy link

Sunnycl commented Mar 12, 2025

训练数据集720条，测试集80条，共2.5h左右，训练参数使用的是默认参数，训练了30轮看曲线，发现过拟合很严重
Image

可否提供一些训练建议，另外选取了epoch1和epoch22的参数进行生成，用zeroshot方法可以正常生成（且音色相似度也会提高），用sft方法则会出现长间断的空白，声音也很奇怪
@aluminumbox
Copy link
Collaborator

sft是指针对单个说话人训练，需要准备一个人音频1h以上

@Sunnycl
Copy link
Author

Sunnycl commented Mar 13, 2025

sft是指针对单个说话人训练，需要准备一个人音频1h以上

2.5h左右音频是单个人的，来源是B站易中天演讲，人耳听音频质量OK

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@aluminumbox @Sunnycl