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.

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

微调数据集策略(dataset make confuse) #908

Open
EasonQYS opened this issue Aug 22, 2024 · 0 comments
Open

微调数据集策略(dataset make confuse) #908

EasonQYS opened this issue Aug 22, 2024 · 0 comments

Comments

@EasonQYS
Copy link

目标

我希望从主播直播间抓取直播风格的语料数据进行训练。主要是风格迁移,知识能力保持。

做法

目前尝试的是internlm2.5-7b-chat。

制作conversation对,上下文分别放在input和output中,原本500条数据,增广数据(转写法)最终3000条数据。

疑问

但是看到Chat-嬛嬛 2.0(https://github.com/KMnO4-zx/huanhuan-chat )是放在 system中的,不太明白有什么区别(跟system_prompt有不同吗?)。也有看到 chat-沪语-阿宝(https://github.com/hoo01/chat-huyu-ABao )在system中设置的是“阿宝”,input是上文,output是下文。不太明白这些做法的差异。

效果

目前我的效果是:template训练400iter,可以有较好的风格迁移,但是知识能力有丢失,说话很短。然后4bit量化之后,会出现重复说话不停止的问题。

我的demo

数据获取:https://openxlab.org.cn/apps/detail/yisheng/ASR-funasr
7b模型体验:https://openxlab.org.cn/apps/detail/yisheng/steamlit-gpu
4bit版:https://openxlab.org.cn/apps/detail/yisheng/snh48_4bit

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

1 participant