intel/auto-round

Pull Requests Commits

xin3he committed 55 days ago

93ef257a

add consideration for moe

xin3he committed 55 days ago

4e0983cc

add is_memory_reached_threshold check

xin3he committed 55 days ago

8b2e1504

xin3he committed 55 days ago

42b1453a

consider output_device when setting device_map

xin3he committed 56 days ago

488eec1a

fix bug and refine additional memory calcu logic

xin3he committed 56 days ago

135339dc

clear_memory only xpu and remove block.to(device)

xin3he committed 56 days ago

40d634cb

update per review comments

xin3he committed 56 days ago

649ffee7

refine device map logic

xin3he committed 56 days ago

cd5f6851

clear_memory for XPU

xin3he committed 56 days ago

6508b74a

consider enable_act_quant and optimize device map logic

xin3he committed 57 days ago

7c598baf

remove mem_per_param_scale

xin3he committed 57 days ago

a582d466

enhance auto device map and support XPU

xin3he committed 57 days ago

f0cc1315

add self attribution and fix avg_bits error (#956)

xin3he committed 57 days ago

Verified 12c49846

Reduce AutoSchem VRAM usage by up to 10X (#944)

wenhuach21 committed 57 days ago

Verified 90c2fb4c

update gguf and support for CompressedLinear (#950)

n1ck-guo committed 57 days ago

Verified fbb9c13b

update readme for sglang support (#953)

WeiweiZhang1 committed 58 days ago

Verified 824a21fb

refactor utils file (#943)

n1ck-guo committed 58 days ago

Verified f1b5c72b

update readme (#949)

wenhuach21 committed 58 days ago

Verified 959ac67c

Add diffusion README (#923)

mengniwang95 committed 59 days ago

Verified 97512a46

support attention mask in user's dataset (#930)

wenhuach21 committed 59 days ago

Verified dfbbba4f

gguf weight type align with original, output.weight, token_embed (#900)

n1ck-guo committed 59 days ago

Verified af8708e5

Add commit hash to version (#941)

XuehaoSun committed 59 days ago

Verified 9052e31b

Add compatibility test (#918)

XuehaoSun committed 59 days ago

Verified 045a27df

remove low_cpu_mem (#934)

n1ck-guo committed 59 days ago

Verified 1355311b

gguf support for inclusionAI/Ling-flash-2.0 (#940)

n1ck-guo committed 62 days ago

Verified 90a3e861

Fix Flux tuning issue (#936)

mengniwang95 committed 63 days ago

Verified 22bd6b08

enable regex quantization config saving for mixed bits (#825)

WeiweiZhang1 committed 63 days ago

Verified 873114a0

remove numpy<2.0 limitation (#921)

xin3he committed 64 days ago

Verified a00fcc60

fix bug of gguf and support for LiquidAI/LFM2-1.2B (#927)

n1ck-guo committed 64 days ago

Verified 80c8de57

Older