hadriansecurity / hadriansecurity/subwiz

Train code + data

Open
#35 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
391
Forks
25
PR merge metrics
No merged PRs in 30d

Description

Guys, are you planning to opensource the training code/data? :)

- I think it makes sense to update the architecture: add RoPE, RMSNorm, qk-norm, etc.
- Experiments with loss, include post training, for example RLVR (reward the model for generating correct domains).

Contributor guide

Open the contributing guide

Assessment

This issue has not been assessed yet.

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.