intel/auto-round

Pull Requests Commits

fix contiguous issue (#1594)

xin3he committed 1 day ago

Verified 4689e93e

Revert "add rtn quant func (#1564)" (#1593)

mengniwang95 committed 1 day ago

Verified d02b2ed4

fix packing nvfp/mxfp `max_wokers` & extend xpu ut (#1555)

Kaihui-intel committed 3 days ago

Verified 79fa1a9b

Support GLM-Image model quantizaiton (#1512)

lvliang-intel committed 3 days ago

Verified 62f92d1f

Fix CI test FP8 static backend matching failure (#1589)

lvliang-intel committed 3 days ago

Verified 6bfe88fb

Add Performance Test (#1561)

XuehaoSun committed 4 days ago

Verified 29f02751

support hadamard transform for mxfp4 with rtn or autoround method (#1515)

lkk12014402 committed 4 days ago

Verified a49550bc

Enable CUDA CI (#1473)

XuehaoSun committed 4 days ago

Verified 4f1a8dea

update awq backend infos, refine UT (#1581)

WeiweiZhang1 committed 4 days ago

Verified 06dfc838

disable torch compile for gguf with alg_ext (#1582)

wenhuach21 committed 4 days ago

Verified fcb0d2d1

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

xin3he committed 4 days ago

Verified 38c70355

Fix compatibility with latest compressed_tensors compressor refactor (#1576)

yiliu30 committed 4 days ago

Verified 49d2dbb0

add rtn quant func (#1564)

mengniwang95 committed 4 days ago

Verified 1ff8d028

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

lkk12014402 committed 4 days ago

Verified 093324c1

Update document for Qwen Omni quantized models (#1568)

lvliang-intel committed 4 days ago

Verified 92f0aaa3

Refine CPU RAM offload strategy (#1544)

lvliang-intel committed 5 days ago

Verified c37add63

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

WeiweiZhang1 committed 5 days ago

Verified 84448965

Refine model save folder name (#1549)

Kaihui-intel committed 5 days ago

Verified 68d5568b

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

xin3he committed 5 days ago

Verified 13fbffee

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

chensuyue committed 5 days ago

Verified 4cc0cce4

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

chensuyue committed 6 days ago

Verified 3a8c26e6

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

lvliang-intel committed 6 days ago

Verified 91af35ff

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

yiliu30 committed 6 days ago

Verified 555052e7

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

xin3he committed 6 days ago

Verified 6d17bbfb

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

lvliang-intel committed 7 days ago

Verified 81e79a14

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

xin3he committed 7 days ago

Verified 1111bcd4

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

yiliu30 committed 7 days ago

Verified 803912aa

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

xin3he committed 7 days ago

Verified d8ee85c8

Support block-wise fp8 quant (#1487)

mengniwang95 committed 7 days ago

Verified e6600591

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

xin3he committed 8 days ago

Verified b5fc1724

Older