debug

Signed-off-by: tzhouam <tzhouam@connect.ust.hk>

vllm_omni/config/model.py

@@ -97,6 +97,7 @@ class OmniModelConfig(ModelConfig):
     def __post_init__(
         self,
         # Multimodal config init vars
+        language_model_only: bool,
         limit_mm_per_prompt: dict[str, int | dict[str, int]] | None,
         enable_mm_embeds: bool | None,
         media_io_kwargs: dict[str, dict[str, Any]] | None,
@@ -113,6 +114,7 @@ class OmniModelConfig(ModelConfig):
     ) -> None:
         # Call parent's __post_init__ to handle all standard ModelConfig initialization
         super().__post_init__(
+            language_model_only=language_model_only,
             limit_mm_per_prompt=limit_mm_per_prompt,
             enable_mm_embeds=enable_mm_embeds,
             media_io_kwargs=media_io_kwargs,

vllm_omni/entrypoints/openai/serving_chat.py

@@ -13,7 +13,7 @@ import torch
 from fastapi import Request
 from PIL import Image
 from pydantic import TypeAdapter
-from vllm.renderers.protocol import BaseRenderer
+from vllm.renderers.base import BaseRenderer
 
 from vllm_omni.entrypoints.async_omni import AsyncOmni
 from vllm_omni.entrypoints.openai.protocol.chat_completion import OmniChatCompletionResponse