intel/auto-round

Pull Requests Commits

[regression] fix copy missing tensors for llama4 (#1579)

xin3he committed 11 days ago

Verified 38c70355

Fix compatibility with latest compressed_tensors compressor refactor (#1576)

yiliu30 committed 12 days ago

Verified 49d2dbb0

add rtn quant func (#1564)

mengniwang95 committed 12 days ago

Verified 1ff8d028

fix deadlock issue when do quantization on hpu. (#1571)

lkk12014402 committed 12 days ago

Verified 093324c1

Update document for Qwen Omni quantized models (#1568)

lvliang-intel committed 12 days ago

Verified 92f0aaa3

Refine CPU RAM offload strategy (#1544)

lvliang-intel committed 12 days ago

Verified c37add63

add gptqmodel:awq backend support and keep autoawq WQLinear_GEMM for compatibility (#1545)

WeiweiZhang1 committed 12 days ago

Verified 84448965

Refine model save folder name (#1549)

Kaihui-intel committed 13 days ago

Verified 68d5568b

Fix reload issue for lm_head quantization and update CI to cover it (#1563)

xin3he committed 13 days ago

Verified 13fbffee

Freeze compressed-tensors & llmc version temporary (#1562)

chensuyue committed 13 days ago

Verified 4cc0cce4

temporary pin compressed-tensors version for llmcompressor (#1560)

chensuyue committed 14 days ago

Verified 3a8c26e6

Support Qwen3 and Qwen2.5 Omni model quantization (#1404)

lvliang-intel committed 14 days ago

Verified 91af35ff

Fix MoE for llama/gpt-oss (#1557)

yiliu30 committed 14 days ago

Verified 555052e7

fix missing change for compress_layer_names and add UTs (#1556)

xin3he committed 14 days ago

Verified 6d17bbfb

Fix shard finalization issue to prevent skipping non-LLM tail layers (#1548)

lvliang-intel committed 14 days ago

Verified 81e79a14

Optimize mtp tensor process by compiling layer names once for all (#1554)

xin3he committed 14 days ago

Verified 1111bcd4

Use AR_WORK_SPACE/offload/ as base dir for offload temp files (#1552)

yiliu30 committed 14 days ago

Verified 803912aa

fix wrong match of ignore_layers and use warning instead of error for mismatch (#1553)

xin3he committed 15 days ago

Verified d8ee85c8

Support block-wise fp8 quant (#1487)

mengniwang95 committed 15 days ago

Verified e6600591

fix google/gemma-3-4b-it (#1547)

xin3he committed 15 days ago

Verified b5fc1724

Add markdown-link-check (#1531)

XuehaoSun committed 16 days ago

Verified 50c4c7f4

fix alg_ext torch compile issue (#1540)

wenhuach21 committed 18 days ago

Verified 9d54ddcc

Fix CI test_transformers issue caused by transformers 5.3.0 (#1535)

lvliang-intel committed 19 days ago

Verified 7b087fb2

Fix sglang ut (#1541)

mengniwang95 committed 19 days ago

Verified 53f2b0e0

support MTP module WOQ quantization and step3p5 model quantization (#1526)

xin3he committed 19 days ago

Verified 42fdb069

fix missing trust_remote_code in unfused_moe (#1532)

xin3he committed 20 days ago

Verified 837888dd

fix dynamic int8 w8a8 export issue with tuning (#1525)

thuang6 committed 20 days ago

Verified a42fd85d

fix broken link (#1530)

xin3he committed 21 days ago

Verified f8e8ae9e

Optimize CPU RAM peak memory during quantization (#1386)

lvliang-intel committed 21 days ago

Verified 0d372151

Support diffusion model saving (#1519)

mengniwang95 committed 21 days ago

Verified a0ad7780

Newer Older