Skip to content

PTQ-Level Efficiency, QAT-Level Accuracy | 训练后量化的效率,量化感知训练的精度

Commits on main

code optimization of vlm (#704)
Heng Guo committed 1 year ago
ef4af0f
3185010
fix bug of zero accuracy for mx-fp (#709)
Heng Guo committed 1 year ago
34051b3
fix xpu oom checker (#705)
Heng Guo committed 1 year ago
32d2cd5
support to save image_processor (#694)
Heng Guo committed 1 year ago
4079a81
Add script for cuda unit test (#567)
Sun, Xuehao committed 1 year ago
4e61755
Update readme (#695)
Wenhua Cheng committed 1 year ago
a598f17
Add black and blacken-docs in pre-commit (#692)
Sun, Xuehao committed 1 year ago
d424d3b
support rceil for mxfp (#660)
Wenhua Cheng committed 1 year ago
bec58a2
better log (#688)
Wenhua Cheng committed 1 year ago
c6bde4b
Add ruff and isort (#578)
Sun, Xuehao committed 1 year ago
774a5ae
support to export static afp8 model (#662)
Heng Guo committed 1 year ago
92bbaad
fix ut import issue (#686)
Weiwei committed 1 year ago
9e47527
bump version into v0.6.0 (#687)
Sun, Xuehao committed 1 year ago
5a07e4a
fix gemma vlm gguf regression (#685)
Wenhua Cheng committed 1 year ago
3f36254
fix gguf regression for large models (#680)
Wenhua Cheng committed 1 year ago
21bb06b
rm_old_vlm_cuda_ut (#678)
Weiwei committed 1 year ago
1ead9fe
rm low_cpu when loading the model (#676)
Wenhua Cheng committed 1 year ago
2fe91d7
308e5e3
support to use lm_eval for vlm (#670)
Heng Guo committed 1 year ago
16b8a26
7863f69
fix ut issue (#668)
Weiwei committed 1 year ago
0a8ebdf
handle zeros in imatrix (#667)
Wenhua Cheng committed 1 year ago
c1ea0d9