auto-round
feat: handle per-tensor FP8 dequantization for Devstral models
#1356
Merged
Merged
feat: handle per-tensor FP8 dequantization for Devstral models #1356
yiliu30
merged
8 commits into
intel:main
from
SwekeR-463:feat/fp8-per-tensor-dequant
feat: handle per-tensor FP8 dequantization for Devstral models
8ad08868
![pre-commit-ci[bot]](https://avatars.githubusercontent.com/in/68672?v=4&s=60)
[pre-commit.ci] auto fixes from pre-commit.com hooks
d43b6f19
Merge branch 'main' into feat/fp8-per-tensor-dequant
982e8866
Merge branch 'main' into feat/fp8-per-tensor-dequant
06f55413
Merge origin/main and resolve conflicts in model.py
9c0121ea
refactor for signature and tests with new transformer version
29c12dad
yiliu30
commented
on 2026-01-31
refactor: rename `dequant_block_fp8_weight` to `_dequant_fp8_linear_w…
2c502915
[pre-commit.ci] auto fixes from pre-commit.com hooks
4d258d27
yiliu30
approved these changes
on 2026-02-01
Login to write a write a comment.
Login via GitHub