Skip to content

Issues: uclaml/SPPO

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

14 Open 6 Closed
14 Open 6 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

Discrepancy in rm_scores Length for Mistral-7B-SPPO Iter1 Dataset
#27 opened Oct 9, 2024 by srzhu97
1
Release training dataset and loss log to help reproduce results
#26 opened Oct 9, 2024 by srzhu97
1
Added a new PR to allow generation on fewer than 8 GPUs
#25 opened Sep 2, 2024 by aman2304
DPO baseline implementation
#22 opened Jul 23, 2024 by yesiam-png
SPPO Implementation on Axolotl!
#21 opened Jul 11, 2024 by kaykyr
Scores and probability calcuations
#15 opened Jul 9, 2024 by namdw
4
What's the package configuration for reproduce SPPO-Gemma-2?
#14 opened Jul 8, 2024 by Jackory
1
Ranking speed & training hyperparameters
#10 opened Jul 3, 2024 by skramer-dev
Is it normal the pipeline start with a huge loss ?
#8 opened Jul 2, 2024 by qy1026
3
Some packages' version are too old
#7 opened Jul 2, 2024 by qy1026
Which version of vllm should be installed
#5 opened Jul 1, 2024 by xinghuang2050
4
ShareGPT appending
#4 opened Jun 30, 2024 by Kquant03
ConnectionError: Couldn't reach 'synthetic_data_llama-3-8b-instruct-sppo-iter3_score' on the Hub (ConnectionError)
#2 opened Jun 27, 2024 by xinghuang2050
2
Is it possible to run llama 3-70B and/or mixtral 8x22b through this process?
#1 opened Jun 26, 2024 by RandomInternetPreson
1
ProTip! Find all open issues with in progress development work with linked:pr.