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

[CustomOp] Register RotaryEmbedding instead of overwrite forward (#2385)

Browse Source 
### What this PR does / why we need it?
Register RotaryEmbedding instead of overwrite forward

### Does this PR introduce _any_ user-facing change?
N/A

### How was this patch tested?
CI passed with new added/existing test.

- vLLM version: v0.10.0
- vLLM main:
808d2e9aa0

---------

Signed-off-by: Icey <1790571317@qq.com>
Signed-off-by: wxsIcey <1790571317@qq.com>
This commit is contained in:
Icey
2025-08-25 09:32:35 +08:00
committed by GitHub
parent 950c4b219a
commit f796e6280b
6 changed files with 426 additions and 381 deletions
Download Patch File Download Diff File Expand all files Collapse all files

7
vllm_ascend/utils.py
View File

@@ -478,9 +478,16 @@ def register_ascend_customop():
from vllm_ascend.ops.linear import (AscendMlpColumnParallelLinear,
AscendMlpMergedColumnParallelLinear,
AscendMlpRowParallelLinear)
from vllm_ascend.ops.rotary_embedding import (
AscendDeepseekScalingRotaryEmbedding, AscendRotaryEmbedding)
CustomOp.register_oot(_decorated_op_cls=AscendQuickGELU, name="QuickGELU")
CustomOp.register_oot(_decorated_op_cls=AscendSiluAndMul,
name="SiluAndMul")
CustomOp.register_oot(_decorated_op_cls=AscendRotaryEmbedding,
name="RotaryEmbedding")
CustomOp.register_oot(
_decorated_op_cls=AscendDeepseekScalingRotaryEmbedding,
name="DeepseekScalingRotaryEmbedding")
if envs_ascend.VLLM_ASCEND_ENABLE_MLP_OPTIMIZE:
CustomOp.register_oot(_decorated_op_cls=AscendMlpColumnParallelLinear,
name="ColumnParallelLinear")