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

EVS Support (Video tokens pruning) (#22980)

Browse Source 
Signed-off-by: Eugene Khvedchenia <ekhvedchenia@nvidia.com>
Signed-off-by: Eugene Khvedchenya <ekhvedchenya@gmail.com>
Co-authored-by: Roger Wang <hey@rogerw.io>
This commit is contained in:
Eugene Khvedchenya
2025-09-26 06:54:54 +03:00
committed by GitHub
parent 983056e456
commit 392edee34a
8 changed files with 783 additions and 39 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
vllm/engine/arg_utils.py
View File

@@ -391,6 +391,7 @@ class EngineArgs:
mm_encoder_tp_mode: MMEncoderTPMode = MultiModalConfig.mm_encoder_tp_mode
io_processor_plugin: Optional[str] = None
skip_mm_profiling: bool = MultiModalConfig.skip_mm_profiling
video_pruning_rate: float = MultiModalConfig.video_pruning_rate
# LoRA fields
enable_lora: bool = False
enable_lora_bias: bool = LoRAConfig.bias_enabled
@@ -813,6 +814,9 @@ class EngineArgs:
multimodal_group.add_argument("--skip-mm-profiling",
**multimodal_kwargs["skip_mm_profiling"])
multimodal_group.add_argument(
"--video-pruning-rate", **multimodal_kwargs["video_pruning_rate"])
# LoRA related configs
lora_kwargs = get_kwargs(LoRAConfig)
lora_group = parser.add_argument_group(
@@ -1032,6 +1036,7 @@ class EngineArgs:
model_impl=self.model_impl,
override_attention_dtype=self.override_attention_dtype,
logits_processors=self.logits_processors,
video_pruning_rate=self.video_pruning_rate,
io_processor_plugin=self.io_processor_plugin,
)