transformers
6247bd14 - Merge branch 'main' into unify-resolve-plan

Commit

9 days ago

Merge branch 'main' into unify-resolve-plan

References

#49429 - Expert-parallel implementations for `ep_dispatch_experts`: all-reduce at inference, all-to-all in training

#49527 - 🚨 build and expand once the tp/ep_plan/fsdp_plan

#49606 - `deepgemm_megamoe` EP implementation for DeepGEMM Mega MoE

#49607 - Move the remaining router-masked MoE plans to `ep_dispatch_experts`

#49613 - 🚨 Run MXFP4 GPT-OSS experts through `ep_dispatch_experts`

#49615 - 🚨 Remove the `ep_router` expert-parallel style

#49638 - Add the `deepep` expert-parallel implementation

Author

3outeille

3outeille

Parents

Loading