Update app.py

app.py

@@ -6,6 +6,10 @@ import faiss
 import torch
 import json
 import spaces
+import logging
+
+# Set up logging
+logging.basicConfig(level=logging.DEBUG)
 
 # Load CSV data
 data = pd.read_csv('RBDx10kstats.csv')
@@ -15,7 +19,7 @@ def safe_json_loads(x):
     try:
         return np.array(json.loads(x), dtype=np.float32)  # Ensure the array is of type float32
     except json.JSONDecodeError as e:
-        print(f"Error decoding JSON: {e}")
+        logging.error(f"Error decoding JSON: {e}")
         return np.array([], dtype=np.float32)  # Return an empty array or handle it as appropriate
 
 # Apply the safe_json_loads function to the embedding column
@@ -59,20 +63,29 @@ def embed_question(question, model, tokenizer):
 # Function to retrieve the relevant document and generate a response
 @spaces.GPU(duration=120)
 def retrieve_and_generate(question):
-    # Embed the question
-    question_embedding = embed_question(question, model, tokenizer)
-    
-    # Search in FAISS index
-    _, indices = gpu_index.search(question_embedding, k=1)
-    
-    # Retrieve the most relevant document
-    relevant_doc = data.iloc[indices[0][0]]
-    
-    # Use the QA model to generate the answer
-    context = relevant_doc['Abstract']
-    response = qa_model(question=question, context=context)
-    
-    return response['answer']
+    logging.debug(f"Received question: {question}")
+    try:
+        # Embed the question
+        question_embedding = embed_question(question, model, tokenizer)
+        logging.debug(f"Question embedding: {question_embedding}")
+
+        # Search in FAISS index
+        _, indices = gpu_index.search(question_embedding, k=1)
+        logging.debug(f"Indices found: {indices}")
+
+        # Retrieve the most relevant document
+        relevant_doc = data.iloc[indices[0][0]]
+        logging.debug(f"Relevant document: {relevant_doc}")
+
+        # Use the QA model to generate the answer
+        context = relevant_doc['Abstract']
+        response = qa_model(question=question, context=context)
+        logging.debug(f"Response: {response}")
+
+        return response['answer']
+    except Exception as e:
+        logging.error(f"Error during retrieval and generation: {e}")
+        return "An error occurred. Please try again."
 
 # Create a Gradio interface
 interface = gr.Interface(