PR #1542 [Step1 ]new architecture for auto_round

init

7698b934

n1ck-guo requested a review from

lkk12014402 98 days ago

n1ck-guo requested a review from

xin3he 98 days ago

n1ck-guo requested a review from

lvliang-intel 98 days ago

n1ck-guo requested a review from

wenhuach21 98 days ago

n1ck-guo requested a review from

copilot-pull-request-reviewer 98 days ago

n1ck-guo added draft

n1ck-guo added engineering

copilot-pull-request-reviewer commented on 2026-03-13

wenhuach21 commented on 2026-03-13

n1ck-guo removed review request from

xin3he 97 days ago

n1ck-guo requested a review from

WeiweiZhang1 97 days ago

n1ck-guo requested a review from

yiliu30 97 days ago

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

75b4141e

update

ca170974

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

a092e379

lvliang-intel commented on 2026-03-16

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

cec4ce4b

chensuyue added this to the 0.12.0 milestone 94 days ago

update

e265b8fe

merge main

868a82db

[pre-commit.ci] auto fixes from pre-commit.com hooks

9dc930c6

add switch

70a2d026

n1ck-guo changed the title ~~[WIP] new architecture for auto_round~~ new architecture for auto_round 93 days ago

wenhuach21 commented on 2026-03-17

n1ck-guo removed draft

n1ck-guo added api/new

code scan

5998d444

n1ck-guo requested a review from

thuang6 93 days ago

n1ck-guo requested a review from

xin3he 93 days ago

n1ck-guo requested a review from

lvliang-intel 93 days ago

Merge branch 'hengguo/new_ar_arch' of https://github.com/intel/auto-r…

94125969

n1ck-guo requested a review from

wenhuach21 93 days ago

yiliu30 commented on 2026-03-17

fix

394dcdd5

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

7024cad0

yiliu30 commented on 2026-03-18

fix

36daba08

fix

6feed993

fix qweight

7bd3e62b

fix ut and refactor code

9b149183

fix ut

2ab9b51b

fix

dd5aec7e

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

d65f1ebf

fix merge

bde95c68

fix

7b4e479a

update

9b4cab71

merge main

b602e008

sync merge change

a1fe717c

fix

b58d55aa

wenhuach21 commented on 2026-03-24

wenhuach21 commented on 2026-03-25

fix ut

6a7ac607

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

64d4a570

decoupling quantization and refactor hadamard

b753bab1

support multi rotation

b32bc685

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

dbd1ab0e

sync compressors_new: add is_dynamic_afp8, is_block_wfp8, _get_safete…

f4da8be2

merge main

75a472a2

n1ck-guo requested a review from

yiliu30 79 days ago

fix

01f68718

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

53bef7c5

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

20ade76c

fix

41e75bd2

fix

92139d6c

wenhuach21 commented on 2026-03-31

fix output dir

166b5b65

lkk12014402 commented on 2026-03-31

wenhuach21 commented on 2026-03-31

update by comment

31b2d2b9

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

4490a17a

wenhuach21 commented on 2026-04-01

wenhuach21 requested changes on 2026-04-01

wenhuach21 requested a review from

wenhuach21 78 days ago

wenhuach21 commented on 2026-04-01

update

fdc92c2f

fix

fb046131

fix by comment

45882798

fix output_dir

a313c264

fix

19f95eda

fix

29d2b64e

merge

bfec8423

lvliang-intel commented on 2026-04-03

fix

1c9e5296

fix vlm ut

7e7fdeb5

lvliang-intel commented on 2026-04-03

[pre-commit.ci] auto fixes from pre-commit.com hooks

4a035fbb

fix ut

463bb6c2

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

f5d6ff4f

sync merge

755ab4e1

fix by comment

d661e0b2

merge

7a80debd

fix

08770cf5

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

709269af

fix

97b89dda

performance

00252569

n1ck-guo added ready

n1ck-guo requested a review from

lkk12014402 71 days ago

n1ck-guo requested a review from

lvliang-intel 71 days ago

fix

18311269

fix

8873eca8

fix

a1a42447

preformance

bd755361

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

f2940bd2

n1ck-guo requested a review from

wenhuach21 69 days ago

sync

e4ce4206

yiliu30 commented on 2026-04-10

fix

1286749c

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

99143064

performance

5c212b56

wenhuach21 commented on 2026-04-10

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

550158be

performance

4806d5aa

fix

1f1fbd93

update

e4fdfe6d

fix: skip compile_func for FP8_STATIC on HPU + trim malloc at ModelCo…

ec45a1c5

fix(memory): reduce peak RSS for new arch via forced malloc_trim and …

3cd3c739

xin3he commented on 2026-04-13

merge main

5d4a85db

fix(memory): reduce peak RAM via deferred ShardWriter, intermediate G…

14a59db1

lkk12014402 approved these changes on 2026-04-14

fix

29969c88

merge main

654c733b

update

c7f21a75

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

72c04f9d

sync hadamard transform changes from main branch to new architecture

028bb069

fix sglang test: switch OPT to Qwen3-0.6B to avoid fused qkv_proj reg…

451008b5

fix: invalidate compiled block forward cache on block change; guard l…

8bdf054e

Merge origin/main: resolve conflicts in test files

8f580393

sync 8d7bb84c to new arch: enable immediate_saving for nv_fp/mx_fp, f…

d067b534

merge main

8d2f3419

fix ut

cec09580

wenhuach21 requested a review from

wenhuach21 62 days ago

wenhuach21 approved these changes on 2026-04-17

wenhuach21 changed the title ~~new architecture for auto_round~~ [Step1 ]new architecture for auto_round 62 days ago

fix HPU FP8_STATIC peak RAM: disable eager pipeline in new-arch, clea…

459435ca

[pre-commit.ci] auto fixes from pre-commit.com hooks

463f58d2

test: add unit tests for HPU FP8_STATIC eager pipeline guard

6cbb1560

fix ut

9f88982b

Merge branch 'hengguo/new_ar_arch' of https://github.com/intel/auto-r…

50beadcb

Merge branch 'main' into hengguo/new_ar_arch

e21c349b

yiliu30 approved these changes on 2026-04-20

merge main

18ba254d

fix merge

e6de66bb

fix

85740b5c

clean

66c4da12

fix W8A16 VRAM regression: skip block_forward compile for zero-shot path

ab9d972d

Merge main: resolve conflicts, sync calib.py shared_cache_keys fix

53bdb74c

sync rotation/hadamard: handle rotation_config kwarg in AutoRoundComp…

70ed236e

sync: rename HadamardConfig to RotationConfig in new arch transforms

c07ca3f1

refactor: rename algorithms/transforms/hadamard -> rotation

4f534100

refactor: dedupe experimental/transform, re-export from new arch rota…

00fc8006

refactor(rotation): physically migrate inplace+dispatcher to new arch

8c96d9b3

refactor(rotation): convert experimental/transform/apply.py to shim

01f02cf2

refactor(entry): simplify rotation_config kwarg handling in AutoRound…

08caa0d4

feat(rotation): support backend='inplace' in new-arch alg_configs pip…

70631e99

[pre-commit.ci] auto fixes from pre-commit.com hooks

d5c8e9ba

merge and sync main

aa4c5407

clean

4c78670d

clean

e4e025eb

fix(transforms): handle block_size=None in HadamardTransform

e3087956

fix cuda ut

110e9ead

Merge remote-tracking branch 'origin/main' into hengguo/new_ar_arch

baf6bd33

merge and sync main

c97428c0

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

bad859a5

sync: add xpu sdpa patch and AutoScheme VLM support to new arch

441d060f

Merge branch 'main' of https://github.com/intel/auto-round into hengg…

dd7b7f0f

merge main

efe2c74b

fix diffusion ut

26aa1066

diffusion: align new arch with old DiffusionCompressor

9d98b0da

fix

2fe5f032

chensuyue merged 91585c7d into main 51 days ago

chensuyue deleted the hengguo/new_ar_arch branch 51 days ago

auto-round
[Step1 ]new architecture for auto_round
#1542

Merged

[Step1 ]new architecture for auto_round #1542

auto-round [Step1 ]new architecture for auto_round #1542 Merged

[Step1 ]new architecture for auto_round #1542

auto-round
[Step1 ]new architecture for auto_round
#1542

Merged