Update app.py

app.py

@@ -20,7 +20,7 @@ from langchain.chains import RetrievalQA
 import json
 
 # Initialize models and clients
-MODEL = 'llama3-groq-70b-8192-tool-use-preview'
+MODEL = 'llama-3.1-70b-versatile'
 client = Groq(api_key=os.environ.get("GROQ_API_KEY"))
 
 vqa_model = AutoModel.from_pretrained('openbmb/MiniCPM-V-2', trust_remote_code=True,
@@ -54,8 +54,8 @@ def classify_function(user_prompt):
     You are a function classifier AI assistant. You are given a user input and you need to classify it into one of the following functions:
 
     - `image_generation`: If the user wants to generate an image.
-    - `image_description`: If the user wants to describe an image.
-    - `document_summarization`: If the user wants to summarize a document.
+    - `image_vqa`: If the user wants to ask questions about an image.
+    - `document_qa`: If the user wants to ask questions about a document.
     - `text_to_text`: If the user wants a text-based response.
 
     Respond with a JSON object containing only the chosen function. For example:
@@ -137,6 +137,13 @@ def handle_input(user_prompt, image=None, audio=None, websearch=False, document=
         answer = tavily_client.qna_search(query=user_prompt)
         return answer, None
 
+    # Handle cases with only image or document input
+    if user_prompt is None or user_prompt.strip() == "":
+        if image:
+            user_prompt = "Describe this image"
+        elif document:
+            user_prompt = "Summarize this document"
+
     # Classify user input using LLM
     function = classify_function(user_prompt)