intel/auto-round

Pull Requests Commits

yiliu30 committed 157 days ago

b6d3683c

yiliu30 committed 161 days ago

1c2d725d

yiliu30 committed 161 days ago

13bfaf1c

load static w8a8

yiliu30 committed 161 days ago

d3d8fcf3

n1ck-guo committed 178 days ago

03aa3976

yiliu30 committed 181 days ago

60c7194f

Merge branch 'export-afp8' into fp8-cache-based-export

yiliu30 committed 181 days ago

cff43a8b

add fp8 kvcache support

yiliu30 committed 181 days ago

eed46a63

fix cuda hanging issue during packing (#669)

WeiweiZhang1 committed 182 days ago

Verified 7863f69d

fix ut issue (#668)

WeiweiZhang1 committed 182 days ago

Verified 0a8ebdfb

handle zeros in imatrix (#667)

wenhuach21 committed 182 days ago

Verified c1ea0d98

Merge branch 'hengguo/export_static_fp8' of https://github.com/intel/auto-round into hengguo/export_static_fp8

n1ck-guo committed 182 days ago

711282a2

n1ck-guo committed 182 days ago

b51895d6

Merge branch 'main' into hengguo/export_static_fp8

n1ck-guo committed 182 days ago

Verified 58d57ddb

better file name for group_size= 0 or -1

n1ck-guo committed 182 days ago

d87b3ce0

n1ck-guo committed 182 days ago

b212e4ad

n1ck-guo committed 182 days ago

ab701845

fix bugs of export deepseek gguf format when iters=0 and q3k accuracy (#665)

n1ck-guo committed 182 days ago

Verified 923683ae

fix gguf block-wise issue (#664)

wenhuach21 committed 183 days ago

Verified 313f436e

fix bug of block-wise quant imatrix (#663)

n1ck-guo committed 183 days ago

Verified 5d68aeb7

support to export static afp8 model

n1ck-guo committed 183 days ago

4dd6e6b6

support vlm models for gguf format

n1ck-guo committed 183 days ago

Verified a2f3f016

fix bug of q4_0 and q5_0 at iters==0 (#658)

n1ck-guo committed 188 days ago

Verified 750a5c10

fix some issues (#655)

wenhuach21 committed 189 days ago

Verified 2f22e29a

support two more calib datasets and fix embedding layer bug (#653)

wenhuach21 committed 189 days ago

Verified 7d724036

speedup mxfp tuning and fix nvfp bug (#647)

wenhuach21 committed 190 days ago

Verified fddcb5d8

Remove cuda UT dependencies version restrictions (#651)

XuehaoSun committed 191 days ago

Verified 35cf154f

Integrate RTN quantization into GGUF packing to enhance robustness (#644)

n1ck-guo committed 191 days ago

Verified 06e0f520

[W4A8]Fix Packing (#648)

yiliu30 committed 192 days ago

Verified a5fa835e

fix norm_bias_tuning (#639)

wenhuach21 committed 192 days ago

Verified 28b01fe3

Older