[Core] [Bugfix] [Multimodal] Fix multimodal profiling and generation for SFT/PTQed models (#20058)

Signed-off-by: Kyle Sayers <kylesayrs@gmail.com>

vllm/model_executor/models/phi4mm.py

@@ -762,6 +762,7 @@ class Phi4MMMultiModalProcessor(BaseMultiModalProcessor[Phi4MMProcessingInfo]):
         prompt: str,
         mm_data: Mapping[str, object],
         mm_kwargs: Mapping[str, object],
+        tok_kwargs: Mapping[str, object],
     ) -> BatchFeature:
         if not mm_data:
             prompt_ids = self.info.get_tokenizer().encode(prompt)
@@ -773,7 +774,7 @@ class Phi4MMMultiModalProcessor(BaseMultiModalProcessor[Phi4MMProcessingInfo]):
             mm_data['audios'] = [(data, sr) for data in audio_data]
 
         processed_outputs = super()._call_hf_processor(prompt, mm_data,
-                                                       mm_kwargs)
+                                                       mm_kwargs, tok_kwargs)
 
         num_img_tokens = [
             self.info.get_num_image_tokens(image_width=img_size[0],