Any plans to provide a slime-based training workflow?
Open
- Dominant language
- Python
- Stars
- 147
- Forks
- 23
- PR merge metrics
- No merged PRs in 30d
Description
Hi SETA team, thanks for the great work on scaling terminal agent environments!
I noticed the current training pipeline under training/tbench_areal_workflow/ is built on AReaL. I'm wondering if there are any plans to also provide a workflow based on https://github.com/THUDM/slime (the Megatron + SGLang + Ray RL framework from THUDM)?
Thanks!
Contributor guide
No contributing guide indexed for this repository
Assessment
This issue has not been assessed yet.