Update app.py

app.py

@@ -20,6 +20,7 @@ from transformers import BitsAndBytesConfig
 import torch
 from huggingface_hub import InferenceClient
 from transformers import AutoTokenizer
+from transformers import AutoImageProcessor
 import spaces 
 IS_SPACES_ZERO = os.environ.get("SPACES_ZERO_GPU", "0") == "1"
 IS_SPACE = os.environ.get("SPACE_ID", None) is not None
@@ -43,11 +44,14 @@ quantization_config = BitsAndBytesConfig(
 
 # Load the tokenizer associated with your 'MODEL_ID'
 tokenizer_image_to_text = AutoTokenizer.from_pretrained(MODEL_ID)
+# Load the image processor associated with your 'MODEL_ID'
+image_processor = AutoImageProcessor.from_pretrained(MODEL_ID) 
 # Load models only once
 processor = AutoProcessor.from_pretrained(MODEL_ID)
 model = LlavaForConditionalGeneration.from_pretrained(MODEL_ID, quantization_config=quantization_config, device_map="auto")
 # Pass the tokenizer explicitly to the pipeline
-pipe_image_to_text = pipeline("image-to-text", model=model, tokenizer=tokenizer_image_to_text, model_kwargs={"quantization_config": quantization_config})
+# Pass the image processor explicitly to the pipeline
+pipe_image_to_text = pipeline("image-to-text", model=model, tokenizer=tokenizer_image_to_text, image_processor=image_processor, model_kwargs={"quantization_config": quantization_config})
 # Initialize the text generation pipeline
 pipe_text = pipeline("text-generation", model=TEXT_MODEL_ID, model_kwargs={"quantization_config": quantization_config})