biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

support torch.compile for bailing moe (#21664)

Browse Source
This commit is contained in:
Jinzhen Lin
2025-07-27 07:54:32 +08:00
committed by GitHub
parent 6c66f28fa5
commit c657369841
1 changed files with 2 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
vllm/model_executor/models/bailing_moe.py
View File

@@ -32,6 +32,7 @@ from torch import nn
from transformers.configuration_utils import PretrainedConfig from transformers.configuration_utils import PretrainedConfig
from vllm.attention import Attention from vllm.attention import Attention
from vllm.compilation.decorators import support_torch_compile
from vllm.config import CacheConfig, VllmConfig from vllm.config import CacheConfig, VllmConfig
from vllm.distributed import (get_pp_group, get_tensor_model_parallel_rank, from vllm.distributed import (get_pp_group, get_tensor_model_parallel_rank,
get_tensor_model_parallel_world_size, get_tensor_model_parallel_world_size,
@@ -291,6 +292,7 @@ class BailingMoeBlock(nn.Module):
return hidden_states, residual return hidden_states, residual
@support_torch_compile
class BailingMoeModel(nn.Module): class BailingMoeModel(nn.Module):
def __init__( def __init__(