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

[VLM] Remove BaseProcessingInfo.get_mm_max_tokens_per_item (#16408)

Browse Source 
Signed-off-by: DarkLight1337 <tlleungac@connect.ust.hk>
This commit is contained in:
Cyrus Leung
2025-04-11 00:06:58 +08:00
committed by GitHub
parent 7678fcd5b6
commit 83b824c8b4
39 changed files with 104 additions and 677 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
vllm/model_executor/models/mllama4.py
View File

@@ -498,17 +498,6 @@ class Mllama4ProcessingInfo(BaseProcessingInfo):
image_processor = self.get_hf_processor().image_processor
return image_processor.max_patches
def get_mm_max_tokens_per_item(
self,
seq_len: int,
mm_counts: Mapping[str, int],
) -> Mapping[str, int]:
vision_config = self.get_hf_config().vision_config
patch_per_chunk = self.get_patch_per_chunk(vision_config)
num_patches = self.get_max_num_tiles() + 1
return {"image": patch_per_chunk * num_patches}
def get_image_size_with_most_features(self) -> ImageSize:
vision_config = self.get_hf_config().vision_config
image_size = vision_config.image_size
@@ -516,14 +505,6 @@ class Mllama4ProcessingInfo(BaseProcessingInfo):
return ImageSize(height=self.get_max_num_tiles() * image_size,
width=image_size)
def get_max_image_tokens(self) -> int:
target_width, target_height = self.get_image_size_with_most_features()
return self.get_num_image_tokens(
image_width=target_width,
image_height=target_height,
)
class Mllama4MultiModalProcessor(BaseMultiModalProcessor[Mllama4ProcessingInfo]
):