Releases: silveroxides/convert_to_quant
Releases · silveroxides/convert_to_quant
v1.2.8
What's Changed
- Feature/further optimizations 1 by @silveroxides in #43
- Fix/prodigy convrot standstill by @silveroxides in #44
Full Changelog: v1.2.7...v1.2.8
v1.2.7
v1.2.6
Full Changelog: v1.2.5...v1.2.6
v1.2.5
What's Changed
- Add quantize function as a callable function via imported module by @silveroxides in #40
Full Changelog: v1.2.4...v1.2.5
v1.2.4
Full Changelog: v1.2.3...v1.2.4
v1.2.3
What's Changed
- Fix issue with learned rounding not calculating correctly for int8 ro… by @silveroxides in #39
Full Changelog: v1.2.2...v1.2.3
v1.2.2
What's Changed
- fix int8 rowwise with original optimizer by @silveroxides in #38
Full Changelog: v1.2.1...v1.2.2
v1.2.1
What's Changed
- Add support for row-wise int8 quantization using fused triton kernel by @silveroxides in #36
Full Changelog: v1.2.0...v1.2.1
v1.2.0
What's Changed
- Dev clean branch reformat by @silveroxides in #33
- add yapf for formatting using better looking clang-format based styling by @silveroxides in #34
- Merge commits for formatting and cleanup by @silveroxides in #35
Full Changelog: v1.1.5...v1.2.0
v1.1.5
What's Changed
- add generic text encoder flag to trigger input scales addition by @silveroxides in #30
Full Changelog: v1.1.4...v1.1.5