List view

[not for land] [mxfp8 training] fix TP bug

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

module: trainingquantize_ api training flowquantize_ api training flow

mx

Status: Open (in progress).

pytorch/aonumber 3985

#3985 In pytorch/ao;

· danielvegamyhre opened on Mar 4, 2026

[ROCm] Enable MXFP8/MXFP4 emulation tests on ROCm (MI300+)

ciflow/rocm

ciflow/rocm-mi300

CLA SignedThis label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed.

module: rocm

module: trainingquantize_ api training flowquantize_ api training flow

Status: Open (in progress).

pytorch/aonumber 4041

#4041 In pytorch/ao;

· brucechanglongxu opened on Mar 10, 2026

Triton + Symmetric Memory fused all2all dispatch + token group padding kernel

moe

mx

Status: Open.

#4050 In pytorch/ao;

· danielvegamyhre opened on Mar 11, 2026

CUDA -> CuteDSL 3d weight quantization kernel for backward dgrad computation with RCEIL scaling

moe

mx

Status: Open.

#4051 In pytorch/ao;

· danielvegamyhre opened on Mar 11, 2026

Triton -> CuteDSL dim0 quantization kernel with RCEIL scaling that writes scales directly to ((32,4),4) layout for tcgen05 mma

mx

Status: Open.

#4052 In pytorch/ao;

· danielvegamyhre opened on Mar 11, 2026

CUDA -> CuteDSL dim1 quantization kernel with RCEIL scaling that writes scales directly to ((32,4),4) layout for tcgen05 mma

mx

Status: Open.

#4053 In pytorch/ao;

· danielvegamyhre opened on Mar 11, 2026

CUDA -> Helion exploration for token group padding/unpadding kernels

Status: Open.

#4062 In pytorch/ao;

· danielvegamyhre opened on Mar 11, 2026