Skip to content

Some improvement on [Reasoning] #53

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
Draft
YanSong97 wants to merge 5 commits into
base: main
Choose a base branch
from
Draft

Some improvement on [Reasoning] #53

YanSong97 wants to merge 5 commits into from

Conversation

@YanSong97
Copy link
Collaborator

@YanSong97 YanSong97 commented Nov 12, 2024 •
edited
Loading

  1. Add LLM-as-Judge tools
  2. Configurate data loading, model resources;
  3. add huggingface model worker;

@YanSong97
1. Add LLM-as-Judge toolkits;
3bf4180 
2. Configurate LM/RM address;
3. Add tensor parallel example;
4. Black format;
@YanSong97
1. add LM lora config;
a475caa 
2. refactor prompt resources, data loading;
3. add genrm infer fn and separate Q&A before value inference
4. batch limit for vllm request
@YanSong97
1. refactor data loading and task;
345cd1d 
2. add lm stop str to cfg file;
3. llm-as-judge also extract clean answer for evaluation;
4. add huggingface model worker;
5. update scripts
@YanSong97 YanSong97 force-pushed the reason_llm_as_judge branch from 2b06d58 to 345cd1d Compare December 8, 2024 21:03
@YanSong97 YanSong97 requested a review from ziyuwan December 8, 2024 21:19
@YanSong97
1. add gsm8k
2e9cd35 
2. add offline RM evaluation script
ziyuwan
ziyuwan reviewed Dec 9, 2024
View reviewed changes
@YanSong97
1. relocate LM, RM name to config;
e6af10f 
2. change file type to yaml;
3. fix rstar data loading
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

@ziyuwan ziyuwan ziyuwan left review comments

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

3 participants

@YanSong97 @ziyuwan