Skip to content

GitLab

Commit 3280b0a0 authored by zhuwenwen's avatar zhuwenwen
Browse files

skip aiter

parent 6df19df0
Show whitespace changes
Inline Side-by-side
Showing with 22 additions and 22 deletions
vllm/v1/spec_decode/eagle.py
View file @ 3280b0a0
...@@ -167,34 +167,34 @@ class EagleProposer: ...@@ -167,34 +167,34 @@ class EagleProposer:
# Determine allowed attention backends once during initialization. # Determine allowed attention backends once during initialization.
self.allowed_attn_types: tuple | None = None self.allowed_attn_types: tuple | None = None
if current_platform.is_rocm(): # if current_platform.is_rocm():
from vllm.v1.attention.backends.rocm_attn import RocmAttentionMetadata # from vllm.v1.attention.backends.rocm_attn import RocmAttentionMetadata
rocm_types = [ # rocm_types = [
TritonAttentionMetadata, # TritonAttentionMetadata,
RocmAttentionMetadata, # RocmAttentionMetadata,
] # ]
# ROCM_AITER_FA is an optional backend # # ROCM_AITER_FA is an optional backend
if find_spec( # if find_spec(
AttentionBackendEnum.ROCM_AITER_FA.get_path(include_classname=False) # AttentionBackendEnum.ROCM_AITER_FA.get_path(include_classname=False)
): # ):
from vllm.v1.attention.backends.rocm_aiter_fa import ( # from vllm.v1.attention.backends.rocm_aiter_fa import (
AiterFlashAttentionMetadata, # AiterFlashAttentionMetadata,
) # )
rocm_types.append(AiterFlashAttentionMetadata) # rocm_types.append(AiterFlashAttentionMetadata)
# TRITON_MLA backend support for MLA models (e.g., DeepSeek) # # TRITON_MLA backend support for MLA models (e.g., DeepSeek)
from vllm.v1.attention.backends.mla.common import MLACommonMetadata # from vllm.v1.attention.backends.mla.common import MLACommonMetadata
rocm_types.append(MLACommonMetadata) # rocm_types.append(MLACommonMetadata)
# FlexAttention backend support # # FlexAttention backend support
from vllm.v1.attention.backends.flex_attention import FlexAttentionMetadata # from vllm.v1.attention.backends.flex_attention import FlexAttentionMetadata
rocm_types.append(FlexAttentionMetadata) # rocm_types.append(FlexAttentionMetadata)
self.allowed_attn_types = tuple(rocm_types) # self.allowed_attn_types = tuple(rocm_types)
# Parse the speculative token tree. # Parse the speculative token tree.
spec_token_tree = self.speculative_config.speculative_token_tree spec_token_tree = self.speculative_config.speculative_token_tree
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment