Commit Graph

6 Commits

Author SHA1 Message Date
Wing Lian
c50c4acbf4 EBFT: Matching Features, Not Tokens: Energy-Based Fine-Tuning of Language Models (#3527) [skip ci]
* EBFT wip

* fixes

* more fixeS

* add missing strided module

* ebft fixes for multi-turn

* make ebft work with async

* add example for ebft w qwen3.5

* fix for split thinking and update yaml for lora over linear attention only

* enforce_eager for vllm arg in schema

* fix sync weights

* fix multi-gpu

* handle updated sig for mm

* ddp fixes

* improve multi-gpu handling, don't calculate logits, adaptive completion length

* chore: lint

* chore: lint

* support completion_mean

* Address corereview feedback

* clamp min IS ratio

* Address PR code review

* more fixes identified

* address code review

* Fix property from rebase conflict
2026-03-24 18:43:46 -04:00
Wing Lian
d8a05744d7 Reverts commits 79908b3c6, 083c5a042, e1ff75624, ff77fa248. (#3496)
The non-root user approach had multiple issues with RunPod
compatibility, sudo PATH handling, and tmux in exec sessions.
Restoring root as the default user for now.
2026-03-13 11:54:09 -04:00
Wing Lian
e1ff756245 become the ubuntu user when root logs in (#3494) 2026-03-13 09:06:54 -04:00
Wing Lian
79908b3c6e use ubuntu user instead of root for uv docker images (#3491) 2026-03-12 20:41:13 -04:00
Wing Lian
77828d3559 uv cloud image should use uv w pip (#3449) 2026-03-02 16:39:26 -05:00
Wing Lian
18f26c19ef add uv axolotl builds (#3431) 2026-02-25 14:46:02 -05:00