add other non-integer scheme support for awq #1892
WeiweiZhang1
merged
142 commits into
main
from
weiwei/mxfp_exp
refactor and support for multi algs fusion
d8fc7cc4
Merge branch 'main' into hengguo/refactor_algs
18e8b15d
![pre-commit-ci[bot]](https://avatars.githubusercontent.com/in/68672?v=4&s=60)
[pre-commit.ci] auto fixes from pre-commit.com hooks
4de34b8d
fix bugs
bcf36330
fix bugs
c5c08df3
fix and hanle shared config
92a9723f
fix and hanle shared config
be623ea1
[pre-commit.ci] auto fixes from pre-commit.com hooks
bc6569d4
[pre-commit.ci] auto fixes from pre-commit.com hooks
006374eb
dynamic map checkpoint naming based on model objective. (#1840)
10d8826a
refine/fix gptq format (#1853)
8bce7b04
fix save_quantized log conflict (#1845)
f2a9b1db
merge main
a9fc7dfe
merge main
e70aeeec
Merge branch 'hengguo/refactor_algs' of https://github.com/intel/auto…
f0e84833
Merge branch 'hengguo/refactor_algs' of https://github.com/intel/auto…
76bef3af
update doc (#1856)
b5d5c34b
Fix Qwen Omni quantization model issue for long form audio generation…
f7a13934
fix bug of qwen and gguf export (#1846)
502b800e
refactor quarot/spinquant rotation with simplying code. (#1849)
717d252e
Enhance CI workflow (#1858)
243c68a9
Fix gemma4 crash issue during quantizing (#1860)
13b15498
relocate awq
5a71dbfa
relocate awq
d72fd9bc
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
dae72212
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
4e7949cf
fix: preserve FP8 llmc metadata for mxfp8 (#1861)
1b8767f1
Fix SDPA bug (#1862)
378831f0
Fix special-model predefined ignore layer filtering (#1863)
75147630
Fix packing format in quantization config and update variable assignm…
3997a8c4
refactor scheme and entry
533f12bb
refactor scheme and entry
70611ec9
modify by comments
6a6f97af
modify by comments
a2ea6119
merge main
d3a391c9
merge main
25956b20
Fix unsupported dtype (#1868)
473d9035
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
ecefed98
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
b3294eda
fix ut
cab89be1
fix ut
93caa5cd
fix
8b2af76b
fix
77a26e6f
Fix HPU CI (#1870)
443f288c
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
6f3abc7d
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
2b26e4f2
add llmc api
0ea57740
add llmc api
f753c062
fix segmentation failure (#1871)
814c6683
Bump build version to v0.14.0 (#1875)
dc269095
Fix SGL CI dependency conflict (#1877)
e6b631b6
gguf format add support for mtp quantization (#1866)
db6b76dd
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
9c7853f4
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
b81643a1
fix gguf error
4c5c4703
fix gguf error
ee10d643
[ARK] Add bf16 + NHD layout support, refactor sage_dynamic_quant (#1882)
c74773ea
Suppress misleading warning when detecting model type for GGUF export…
050c501b
fix ut
8e124610
fix ut
d2ebbd06
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
c4706014
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
4fc9724a
fix performance regression (#1886)
9f254fbe
fix
8e4fd0e0
fix
5a968849
Merge branch 'main' into hengguo/refactor_algs
1fdb43a7
Merge branch 'main' into hengguo/refactor_algs
2638c1c7
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
40a67fa0
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
42d77710
Merge branch 'hengguo/refactor_algs' of https://github.com/intel/auto…
b4ae543f
Merge branch 'hengguo/refactor_algs' of https://github.com/intel/auto…
fde14a58
add mxfp support for awq
5c5caccd
fixtypo
52f68404
[pre-commit.ci] auto fixes from pre-commit.com hooks
a4b54050
clean and update
9edf0d03
refactor pipeline
7fb0839e
change num_samples to property
30f1dd05
fix random rotation and update rotation doc. (#1884)
6afd14c2
Update auto-round-lib release package build (#1895)
0de1eb05
[step 1]refine code to support all devices in torch and hot fix for g…
2794a6e2
Fix CI coverage & bug grep issue (#1893)
5ed21d3e
feat: add MXFP4/MXFP8 quantization support (llmc_compressor format) a…
6cdb2a20
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
2a12b197
fix
eac3ab8d
Fix slow startup time of pytest coverage for unit tests (#1899)
2890673c
feat: improve review-pr skill score from 76% to 90% (#1901)
205e5f60
update llama-cpp-python installation for CUDA CI (#1907)
fb9a772f
fix gguf opt-rtn regression (#1905)
bfa795ca
fix: guard zero-division in GGUF quant kernels to avoid NaN block sca…
d6153cb7
fallback compute type on b70 if needed (#1904)
83cbe978
[ARK] update README (#1906)
0068522d
fix inplace rotation issue (#1903)
d13b1ddd
fix awq cuda CI (#1912)
be67ef3a
Merge remote-tracking branch 'origin/main' into hengguo/refactor_algs
8c731933
add type annotation
241cb393
performance
12814d5f
performance
3fecf0bc
Merge branch 'hengguo/refactor_algs' of https://github.com/intel/auto…
6f219c60
refactor and support for multi algs fusion
ab2eb686
fix
379b6da1
fix import
d0232da9
Merge remote-tracking branch 'origin/hengguo/refactor_algs' into weiw…
3755e8f4
Re-apply AWQ weight-clip work atop refactor_algs; fix disable_opt_rtn…
7622230e
WeiweiZhang1
changed the base branch from
hengguo/refactor_algs
to
hengguo/refactor_algs_clean
39 days ago
refine smooth logic, align with refactor
ae3e1c99
Merge remote-tracking branch 'origin/weiwei/mxfp_exp' into weiwei/mxf…
0f0cd09d
add test for use_v2_mx_scale_search
1aefd31f
remove dead _alg_cls/scale_dtype; AWQ inherits opt_rtn from block qua…
5275d7ce
Merge remote-tracking branch 'origin/hengguo/refactor_algs_clean' int…
2e16cd2c
[pre-commit.ci] auto fixes from pre-commit.com hooks
512a4e55
fix cuda ut
f409d961
fix ut
0abcc0bf
add top import of all alg configs
38482d51
chore: add .worktrees/ to .gitignore
c079f9cc
update by comments
40931857
unify blockio output include diffusion
a716a58c
fix macos device
9aa3ec41
Delete .gitignore
64602d92
Revert "Delete .gitignore"
e391d1c4
update
1346d0b7
fixtypo
a352dd5a
Merge branch 'hengguo/refactor_algs_clean' into weiwei/mxfp_exp
42a09b72
Base automatically changed from
hengguo/refactor_algs_clean
to
main
37 days ago
refine v2 search init_scale
fae4b1da
fix conflict
7f899074
fixtypo
21f6dcb7
Merge branch 'main' into weiwei/mxfp_exp
8528b912
wenhuach21
commented
on 2026-06-23
wenhuach21
commented
on 2026-06-23
wenhuach21
commented
on 2026-06-23
wenhuach21
commented
on 2026-06-23
wenhuach21
commented
on 2026-06-23
fixtypo, rm try/except for awq scale search
fe2974b1
Merge branch 'main' into weiwei/mxfp_exp
a63fdd12
WeiweiZhang1
changed the title
decouple awqtransform, refine shceme check
c428982b
Merge branch 'main' into weiwei/mxfp_exp
7283aae5
[pre-commit.ci] auto fixes from pre-commit.com hooks
e4cc2e21
fixtypo
e63232d5
fix CI
aba4ed44
n1ck-guo
commented
on 2026-06-25
n1ck-guo
commented
on 2026-06-25
n1ck-guo
commented
on 2026-06-25
refine API, fixtypo
c083cbaf
[pre-commit.ci] auto fixes from pre-commit.com hooks
533ae4ff
Potential fix for CI test speed
14a9e485
Potential fix for CI
c88a34fe
[pre-commit.ci] auto fixes from pre-commit.com hooks
d0e88c82
fixtypo
14773e51
add more CI
d6aa2fde
n1ck-guo
approved these changes
on 2026-06-25
wenhuach21
commented
on 2026-06-25
wenhuach21
commented
on 2026-06-25
wenhuach21
approved these changes
on 2026-06-25
revert eval device changes, fix ci
6675fc89
upspeed awq CI
02363394
fixtypo
c9a17608
Reviewers
Assignees
No one assigned
Labels
experimental
Milestone
Login to write a write a comment.
Login via GitHub