intel/auto-round

Pull Requests Commits

mengniwang95 committed 35 days ago

4533c8ae

fix tokenizer and update code

mengniwang95 committed 35 days ago

5d1ae160

add save for vllm model quantization

mengniwang95 committed 35 days ago

42351004

mengniwang95 committed 35 days ago

570e44f9

add opt_rtn func

mengniwang95 committed 35 days ago

59d4c5d1

fix input calibration

mengniwang95 committed 35 days ago

5d9a9ad5

update fp8 conversion

mengniwang95 committed 35 days ago

c3a04332

mengniwang95 committed 35 days ago

1750850f

initial support

mengniwang95 committed 35 days ago

8487ae40

Fix attribute bug for diffusion model (#1338)

mengniwang95 committed 35 days ago

Verified a4ee3a00

update release package and bump version (#1327)

chensuyue committed 35 days ago

Verified 00702194

format add check for scheme args (#1332)

n1ck-guo committed 35 days ago

Verified 7f1dee6e

Delay materializing the replaced model weights until quantization (#1307)

yiliu30 committed 35 days ago

Verified d9d20a12

(feat): add support for g2 fp8 on cpu with LUT (#1254)

SwekeR-463 committed 36 days ago

Verified 6190869b

fix tied weights keys bug (#1330)

wenhuach21 committed 37 days ago

Verified 25364d21

Update publication dates in README and add recipe configuration detai… (#1329)

xin3he committed 37 days ago

Verified 46224d4a

[GGUF] using quant_nontext_module to control whether quant vision model (#1317)

n1ck-guo committed 38 days ago

Verified 24e43976

simplify pull_request_template.md (#1328)

xin3he committed 38 days ago

Verified cf37d8e6

add blockwise way in pure rtn to ease module replacements (#1320)

wenhuach21 committed 38 days ago

Verified 57d1a8ac

set predefined ignore layers (#1318)

wenhuach21 committed 38 days ago

Verified 7c397e14

fix low_cpu and set it to True as default (#1288)

wenhuach21 committed 38 days ago

Verified f654e52d

Add issue templates and PR template with required fields (#1309)

Copilot committed 39 days ago

Verified eaf5885a

fix XPU CI hang issue (#1261)

xin3he committed 39 days ago

Verified 57af5b7f

fix bug of format (#1287)

n1ck-guo committed 39 days ago

Verified 5ba31b6c

Enable q_scale calibration for deepseek v2 on Gaudi (#1299)

mengniwang95 committed 39 days ago

Verified d8ed35a8

open source delta loss (#1300)

wenhuach21 committed 40 days ago

Verified 587eadb8

fix ignore layers regression (#1302)

wenhuach21 committed 40 days ago

Verified 3352a830

Remove itrex format (#1301)

Kaihui-intel committed 40 days ago

Verified 46f88606

refine moe modellings to release orginal expert module's ram (#1265)

WeiweiZhang1 committed 42 days ago

Verified e383eea3

Update unit test script (#1290)

XuehaoSun committed 42 days ago

Verified 8e6f3370

Older