intel/auto-round

Pull Requests Commits

add conditional adjustment for step value in compute_mx_v_scale function

xin3he committed 18 days ago

3aa90426

Merge remote-tracking branch 'origin/xinhe/8-25a' into xinhe/8-19

xin3he committed 19 days ago

e15b06e5

increase precision of loss logging across quantization and training processes

xin3he committed 19 days ago

3df657a2

add debug information for tuning

xin3he committed 19 days ago

fa2a08f0

increase precision of loss logging in quantization and training processes

xin3he committed 19 days ago

f1d25130

Update alg combination news (#2227)

wenhuach21 committed 20 days ago

Verified bd9a13ce

fix: update default patience for early stopping in SignRound to 50

xin3he committed 20 days ago

0ad7c578

update auto_scheme_acc doc (#2000)

n1ck-guo committed 20 days ago

Verified 5683ea15

add awq torch_compile fix, disable_opt_rtn by default for awq (#2208)

WeiweiZhang1 committed 20 days ago

Verified 611d4402

update minmax_scale_bound

xin3he committed 20 days ago

b7cbfc32

xin3he committed 20 days ago

97d1cb6e

disable timeout in pytest until found a more stable method (#2222)

chensuyue committed 20 days ago

Verified 96ce4480

feat: add compute_mx_v_scale function for per-element v step-scale in MXFP quantization

xin3he committed 20 days ago

0e51f1b7

Preserve diffusion component dtypes during quantization (#2205)

changwangss committed 20 days ago

Verified d6ead479

refactor: optimize expert selection and output handling in moe forward (#2209)

xin3he committed 23 days ago

Verified b9f3d007

add workaround for llmc issue and add unit tests for model_free_utils helpers (#2207)

xin3he committed 23 days ago

Verified a2c27b01

Fix CT export in mixed‑precision mode (#2128)

1607662794 committed 23 days ago

Verified d034df53

fix: layer_config for lm_head takes effect in model-free mode (#2182)

Copilot committed 24 days ago

Verified 1d050463

feat: add support for FP8 model loading and ragged-tile dequantization fallback (#2158)

xin3he committed 24 days ago

Verified 40635d02

Reduce SYCL-TLA Compiler Memory (#2193)

luoyu-intel committed 24 days ago

Verified ff2fd25c

Increase memory limits for the build process in the CI pipeline (#2198)

chensuyue committed 25 days ago

Verified 8ae2b2e1

update readme (#2194)

wenhuach21 committed 25 days ago

Verified 00213aed

For compatibility with Stable Diffusion (#2191)

changwangss committed 25 days ago

Verified 8884cac8

Fix wan i2v pipeline (#2180)

changwangss committed 25 days ago

Verified 9322f3a3

[Experimental] Add NVFP4 E5M3 support and related components for model-free quantization (#2123)

xin3he committed 25 days ago

Verified 0c0511b5

adjust lrx2 logic to fix its issue in mixed bits scenarios and add more woq support in llmc (#2164)

wenhuach21 committed 25 days ago

Verified 7e35bd63

fix: support aggregated Granite MoE tensors in gguf export (#2175)

n1ck-guo committed 26 days ago

Verified 5a24c84a

Enable Torch compilation for throughput tests (#2181)

xin3he committed 26 days ago

Verified e55beddc

fix DiffusionCompressionOrchestrator missing nsamples (#2179)

changwangss committed 26 days ago

Verified f239f8c9

Simplify default calibration dataset log (#2162)

changwangss committed 26 days ago

Verified 960dc7f8

Older