1
Ask HN: How do you fine-tune LLMs?
Do you usually do full fine-tuning or LoRA (or its variants)? and why?
GitHub: https://github.com/unarxiv/cvpm
中文介绍(Chinese Intro): https://zhuanlan.zhihu.com/p/45672318
CVPM (Computer Vision Package Manager) is an open source software to help developers download, install and run computer vision services. It is the 'pip' for computer vision.
Though it is still in a very early stage and not recommend for use in production (actually, I do not recommend you to try it now because it may contain unknown bugs), We'd like to invite adorable developers to join us. We have a roadmap draft as below:
* Complete the test cases and docs.
* Complete the Model Hub. (Prototype is at https://hub.autoai.org)
* Complete packages for some classic tasks as "Official Repo"
* Complete third party package upload.
As an open source software, I am really sorry that we cannot promise you anything. But the following is what I can do:
* We will have an Authors List for all the contributors.
* There will be logo banner and donors page at the bottom of our Model Hub for sponsors.
If you are interested, please contact me at xiaozhe.yaoi@gmail.com, or you can post a GitHub Issues.