[feat] Tensor parallelism for MiniMax-H3 - #14609

Draft

JingyaHuang wants to merge 12 commits into

huggingface:mainfrom

JingyaHuang:add-h3-tp-support

Commits on Aug 20, 2026

[core] Shard tensor-parallel checkpoints on load and save
JingyaHuang
committed
acdf4bf
Merge branch 'main' into add-shard-ckpt-loading
JingyaHuang
authored
0b9686b

Commits on Aug 21, 2026

Raise when tensor parallelism is combined with quantization, offloading or LoRA
JingyaHuang
committed
40ddb53
Merge branch 'add-shard-ckpt-loading' of github.com:JingyaHuang/diffusers into add-shard-ckpt-loading
JingyaHuang
committed
0760934
Add tensor-parallel support for MiniMax-H3
JingyaHuang
committed
a8956f4
Shard MiniMax-H3's adaln_proj to fit tensor parallelism on one device

JingyaHuang
and
claude
committed
7e6e38f
Build MiniMax-H3's row timestep plan on CPU, as its caller expects

JingyaHuang
and
claude
committed
bb236dc

Commits on Aug 26, 2026

Merge branch 'main' into add-h3-tp-support
JingyaHuang
authored
aa50ad3

Commits on Sep 1, 2026

feat:combine tp+cp
JingyaHuang
committed
0f9057b

Commits on Sep 7, 2026

Allow tensor parallelism and context parallelism in one ParallelConfig

whn09
and
claude
committed
9c81ac6

Commits on Sep 16, 2026