intel/auto-round

Pull Requests Commits

fix bug of q_layer_inputs (#811)

n1ck-guo committed 107 days ago

Verified e1ae6e93

Support loading for static quant weight fp8 act fp8 (#730)

yiliu30 committed 107 days ago

Verified 09e4d312

remove torch compile (#807)

wenhuach21 committed 107 days ago

Verified 1452d95f

fix cuda ut bug of use_deterministic_algorithms (#805)

n1ck-guo committed 107 days ago

Verified 886d6571

fix mxfp exporting (#806)

WeiweiZhang1 committed 107 days ago

Verified 520c78fe

improve the robustness of scheme (#803)

wenhuach21 committed 108 days ago

Verified 9534d2ac

AutoRoundMLLM supports scheme and fix device_map=dict regression (#801)

n1ck-guo committed 108 days ago

Verified 78fce450

enable model python files saving (#802)

WeiweiZhang1 committed 108 days ago

Verified da089e9f

fix several cuda ut bug (#797)

n1ck-guo committed 108 days ago

Verified ab55a973

fix typo in readme (#799)

wenhuach21 committed 108 days ago

Verified e39f25e1

refine readme (#798)

wenhuach21 committed 109 days ago

Verified bc045f93

fix several regressions including lm-head quantization, 3bit asym torch backend,etc (#796)

wenhuach21 committed 109 days ago

Verified 08d6b9e7

add fp8 export format check (#779)

n1ck-guo committed 109 days ago

Verified 5ead4c6d

align formats packing device to API (#795)

WeiweiZhang1 committed 111 days ago

Verified dc81a878

rename llmcompressor to llm_compressor for align with other formats (#780)

WeiweiZhang1 committed 111 days ago

Verified e66652a4

fix bug and add nvfp in alg-ext with slight improvement (#794)

wenhuach21 committed 111 days ago

Verified 7e014ca3

Support auto device mapping (#781)

Kaihui-intel committed 111 days ago

Verified d7d2efad

torch backend bugfix and speedup ut (#793)

WeiweiZhang1 committed 111 days ago

Verified 4c597def

add limit arg for eval (#764)

n1ck-guo committed 112 days ago

Verified add749ac

refine nvfp4 algorithm (#790)

wenhuach21 committed 112 days ago

Verified 536810e1

mxfp/nvfp/fp8 support torch compile in tuning (#789)

wenhuach21 committed 113 days ago

Verified 5f87a06b

refine nvfp code, typofix (#777)

WeiweiZhang1 committed 113 days ago

Verified fced2ba7

fix alg_ext regression and support mxfp4 in it with slight improvement (#785)

wenhuach21 committed 113 days ago

Verified ca3f733e

hot fix for nfp4 scheme (#784)

wenhuach21 committed 114 days ago

Verified 7e52af2a

Speedup FP4 packing (#766)

yiliu30 committed 114 days ago

Verified 253b48f2

[API CHANGE] Stage 1 add quant scheme and consolidate device and device_map (#774)

wenhuach21 committed 114 days ago

Verified 8bcf7eab

gguf format supports for fp8 model (#778)

n1ck-guo committed 115 days ago

Verified 572aad1d

fix bug of auto_round exporting (#772)

n1ck-guo committed 115 days ago

Verified d01704da

fix set_layer_config bug (#768)

wenhuach21 committed 119 days ago

Verified bf58cf97

Keep the model’s buffer dtype unchanged in most cases (#770)

wenhuach21 committed 120 days ago

Verified b2c7a3de

Newer Older