Update app.py

app.py

@@ -4,7 +4,7 @@ import pdfplumber
 from sentence_transformers import SentenceTransformer
 import faiss
 import numpy as np
-from groq import Client
+from transformers import pipeline  # Use Hugging Face model instead of Groq
 
 # Set background image and customize colors
 background_image_url = "https://www.shutterstock.com/image-vector/artificial-intelligence-circuit-electric-line-600nw-2465096659.jpg"
@@ -46,25 +46,29 @@ st.markdown(
         background-color: green;
         color: white;
     }}
-
+    
     /* Set query input block background color to white */
     .stTextInput input {{
         background-color: white;
         color: black;
     }}
 
+    /* Display generated response text in white */
+    .stWrite {{
+        color: white !important;
+    }}
     </style>
     """, 
     unsafe_allow_html=True
 )
 
-# Load Hugging Face Secrets
+# Load Hugging Face Secrets (if needed for Hugging Face integration)
 HUGGINGFACE_KEY = os.getenv("HUGGINGFACE_KEY")
 if not HUGGINGFACE_KEY:
     st.error("Hugging Face API token not found. Please set it in the Hugging Face Secrets.")
 
-# Initialize Groq client
-groq_client = Client(api_key=HUGGINGFACE_KEY)
+# Initialize Hugging Face pipeline for text generation (using GPT-2 or other models)
+generator = pipeline('text-generation', model='gpt-2', api_key=HUGGINGFACE_KEY)
 
 # Load the SentenceTransformer model for embedding generation
 embedder = SentenceTransformer('sentence-transformers/all-MiniLM-L6-v2')
@@ -141,13 +145,13 @@ def process_and_store_document(file_path):
 # User interface for Streamlit
 st.title("The Rise of Agentic AI RAG Application")
 
+# Query input for the user
+user_query = st.text_input("Enter your query:")
+
 # Button to trigger document processing
 if st.button("Process PDF"):
     process_and_store_document(file_path)
 
-# Query input for the user
-user_query = st.text_input("Enter your query:")
-
 if user_query:
     # Check if there are any chunks in the index
     if not chunks:
@@ -177,13 +181,13 @@ if user_query:
                 for chunk in retrieved_chunks:
                     st.write(chunk)
                 
-                # Combine the retrieved chunks with the query and generate a response using Groq
-                combined_input = " ".join(retrieved_chunks) + user_query
-                response = groq_client.generate(model="llama3-8b-8192", prompt=combined_input, max_tokens=200)
+                # Combine the retrieved chunks with the query and generate a response using Hugging Face
+                combined_input = " ".join(retrieved_chunks) + " " + user_query
+                response = generator(combined_input, max_length=200, num_return_sequences=1)[0]['generated_text']
                 
-                # Display the generated response
+                # Display the generated response in white text
                 st.subheader("Generated Response")
-                st.write(response["text"])
+                st.markdown(f"<p style='color:white'>{response}</p>", unsafe_allow_html=True)
 
 # Footer
 st.markdown("<div class='footer'>Created by Shamil Shahbaz</div>", unsafe_allow_html=True)