xinyun/vllm
1
0
Fork 0
mirror of https://git.datalinker.icu/vllm-project/vllm.git synced 2025-12-24 05:55:02 +08:00
Code Issues Packages Projects Releases Wiki Activity

[torch.compile] Add encoder tag for compilation (#30489)

Browse Source 
Signed-off-by: ilmarkov <markovilya197@gmail.com>
This commit is contained in:
Ilya Markov 2025-12-14 11:15:11 +01:00 committed by GitHub
parent 3a20450d31
commit 3224ea9915
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
3 changed files with 14 additions and 10 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

11
vllm/compilation/backends.py
View File

@ -463,21 +463,27 @@ class PiecewiseCompileInterpreter(torch.fx.Interpreter):
# the tag for the part of model being compiled, # the tag for the part of model being compiled,
# e.g. backbone/eagle_head # e.g. backbone/eagle_head
model_tag: str = "backbone" model_tag: str = "backbone"
model_is_encoder: bool = False
@contextmanager @contextmanager
def set_model_tag(tag: str): def set_model_tag(tag: str, is_encoder: bool = False):
"""Context manager to set the model tag.""" """Context manager to set the model tag."""
global model_tag global model_tag
global model_is_encoder
assert tag != model_tag, ( assert tag != model_tag, (
f"Model tag {tag} is the same as the current tag {model_tag}." f"Model tag {tag} is the same as the current tag {model_tag}."
) )
old_tag = model_tag old_tag = model_tag
old_is_encoder = model_is_encoder
model_tag = tag model_tag = tag
model_is_encoder = is_encoder
try: try:
yield yield
finally: finally:
model_tag = old_tag model_tag = old_tag
model_is_encoder = old_is_encoder
class VllmBackend: class VllmBackend:
@ -523,6 +529,9 @@ class VllmBackend:
# them, e.g. backbone (default), eagle_head, etc. # them, e.g. backbone (default), eagle_head, etc.
self.prefix = prefix or model_tag self.prefix = prefix or model_tag
# Mark compilation for encoder.
self.is_encoder = model_is_encoder
# Passes to run on the graph post-grad. # Passes to run on the graph post-grad.
self.pass_manager = resolve_obj_by_qualname( self.pass_manager = resolve_obj_by_qualname(
current_platform.get_pass_manager_cls() current_platform.get_pass_manager_cls()

7
vllm/compilation/piecewise_backend.py
View File

@ -53,12 +53,7 @@ class PiecewiseBackend:
self.is_last_graph = piecewise_compile_index == total_piecewise_compiles - 1 self.is_last_graph = piecewise_compile_index == total_piecewise_compiles - 1
self.is_full_graph = total_piecewise_compiles == 1 self.is_full_graph = total_piecewise_compiles == 1
# TODO: we need to generalize encoder compilation to other models self.is_encoder_compilation = vllm_backend.is_encoder
self.is_encoder_compilation = vllm_backend.prefix in [
"Qwen2_5_VisionPatchEmbed",
"Qwen2_5_VisionPatchMerger",
"Qwen2_5_VisionBlock",
]
self.compile_ranges = self.compilation_config.get_compile_ranges() self.compile_ranges = self.compilation_config.get_compile_ranges()
if self.is_encoder_compilation: if self.is_encoder_compilation:

6
vllm/model_executor/models/qwen2_5_vl.py
View File

@ -612,7 +612,7 @@ class Qwen2_5_VisionTransformer(nn.Module):
# DO NOT MOVE THIS IMPORT # DO NOT MOVE THIS IMPORT
from vllm.compilation.backends import set_model_tag from vllm.compilation.backends import set_model_tag
with set_model_tag("Qwen2_5_VisionPatchEmbed"): with set_model_tag("Qwen2_5_VisionPatchEmbed", is_encoder=True):
self.patch_embed = Qwen2_5_VisionPatchEmbed( self.patch_embed = Qwen2_5_VisionPatchEmbed(
patch_size=patch_size, patch_size=patch_size,
temporal_patch_size=temporal_patch_size, temporal_patch_size=temporal_patch_size,
@ -651,7 +651,7 @@ class Qwen2_5_VisionTransformer(nn.Module):
f"Qwen2.5-VL does not support {self.attn_backend} backend now." f"Qwen2.5-VL does not support {self.attn_backend} backend now."
) )
with set_model_tag("Qwen2_5_VisionBlock"): with set_model_tag("Qwen2_5_VisionBlock", is_encoder=True):
self.blocks = nn.ModuleList( self.blocks = nn.ModuleList(
[ [
Qwen2_5_VisionBlock( Qwen2_5_VisionBlock(
@ -670,7 +670,7 @@ class Qwen2_5_VisionTransformer(nn.Module):
] ]
) )
with set_model_tag("Qwen2_5_VisionPatchMerger"): with set_model_tag("Qwen2_5_VisionPatchMerger", is_encoder=True):
self.merger = Qwen2_5_VisionPatchMerger( self.merger = Qwen2_5_VisionPatchMerger(
d_model=vision_config.out_hidden_size, d_model=vision_config.out_hidden_size,
context_dim=self.hidden_size, context_dim=self.hidden_size,