0.0.22
·
415 commits
to master
since this release
- Fix regression causing models with preserved bf16 tensors (multimodal specifically) to fail quantization
Full Changelog: v0.0.21...v0.0.22