EngineX/xc-llm-ascend
3
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

[MOE Refactor] Remove QuantType in prepare_finalize.py (#6534)

Browse Source 
### What this PR does / why we need it?
To prevent confusion between different QuantType classes, we remove**
QuantType in prepare_finalize.py

- vLLM version: v0.15.0
- vLLM main: https://github.com/vllm-project/vllm/commit/v0.15.0

Signed-off-by: shenchuxiaofugui <1311027364@qq.com>
This commit is contained in:
LI SHENGYONG
2026-02-10 15:59:58 +08:00
committed by GitHub
parent 34eecacace
commit 7cf285a77a
4 changed files with 9 additions and 24 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
tests/ut/ops/test_moe_comm_method.py
View File

@@ -7,7 +7,7 @@ from tests.ut.base import TestBase
from vllm_ascend.ops.fused_moe.moe_comm_method import (AllGatherCommImpl,
AlltoAllCommImpl,
MC2CommImpl)
from vllm_ascend.ops.fused_moe.prepare_finalize import QuantType
from vllm_ascend.quantization.methods.base import QuantType
from vllm_ascend.ops.fused_moe.token_dispatcher import (TokenCombineResult,
TokenDispatchResult)