weichen ffe51eedd6 [Refactor][MoE] Reuse vLLM's all_reduce logic (#5189) ...

### What this PR does / why we need it?
Move all_reduce logic to AscendFusedMoE.forward, reuse vLLM's logic.

### Does this PR introduce _any_ user-facing change?
No

### How was this patch tested?
e2e & ut
- vLLM version: v0.12.0
- vLLM main:
ad32e3e19c

Signed-off-by: weichen <calvin_zhu0210@outlook.com>
Co-authored-by: weijinqian0 <1184188277@qq.com>

2025-12-23 18:53:48 +08:00

..

__init__.py

[Refactor] [MoE] Rename moe-related classes & files (#3646)

2025-10-25 11:22:03 +08:00

comm_utils.py

[Refactor] [MoE] Rename moe-related classes & files (#3646)

2025-10-25 11:22:03 +08:00

experts_selector.py

[refactor] Remove unnecessary attributes from set_ascend_forward_context (#5204)

2025-12-23 08:49:52 +08:00

fused_moe.py

[Refactor][MoE] Reuse vLLM's all_reduce logic (#5189)

2025-12-23 18:53:48 +08:00

moe_comm_method.py

[Feature]Use DispatchGmmCombineDecode operator to replace MC2(Optional) (#5040)

2025-12-21 15:23:59 +08:00

moe_mlp.py

[refactor] Remove unnecessary attributes from set_ascend_forward_context (#5204)

2025-12-23 08:49:52 +08:00

prepare_finalize.py

[Refactor][MoE] Reuse vLLM's all_reduce logic (#5189)

2025-12-23 18:53:48 +08:00

token_dispatcher.py

[Feature]Use DispatchGmmCombineDecode operator to replace MC2(Optional) (#5040)

2025-12-21 15:23:59 +08:00