Add support for MXFP8 All gather #3435

Open

avizon-aws wants to merge 17 commits intopytorch:mainfrom

avizon-aws:mxfp8_ag_feature

Commits on Dec 5, 2025

fixed ruff formatting
avizon-aws
committed
add mxfp8 and nvfp4 to Llama eval scripts (#3394)

vkuzo
authored and
avizon-aws
committed
flip mx inference scaling setting to RCEIL (#3428)

vkuzo
authored and
avizon-aws
committed
add CLAUDE.local.md to gitignore (#3437)

vkuzo
authored and
avizon-aws
committed
bump python version in tutorial ci workflow (#3439)

danielvegamyhre
authored and
avizon-aws
committed
[CPU] Reland qconv fp8 fusion passes (#3433)

authored and
avizon-aws
committed
Int8Tensor migration cleanup (#3407)

jcaip
authored and
avizon-aws
committed
[xpu][test] Port 2 test/dtypes_{floatx, bitpacking} UT files to intel XPU (#3368)

zxd1997066
authored and
avizon-aws
committed
[xpu][test] Port 2 test/quantization/pt2e/test_{quantize_pt2e, quantize_pt2e_qat} UT files to intel XPU (#3405)

zxd1997066
authored and
avizon-aws
committed
[Intel GPU] Enable optim SR test (#3055)

arlesniak
authored and
avizon-aws
committed
updated test with rebase changes
avizon-aws
committed
Merge branch 'pytorch:main' into mxfp8_ag_feature
avizon-aws
authored

Commits on Dec 6, 2025

added checks to run only on CUDA with compatibility >=9
avizon-aws
committed