Update app.py

app.py

@@ -6,11 +6,27 @@ import os
 from sentence_transformers import SentenceTransformer
 from transformers import pipeline
 
-# 1. Load embedding + QA model
+# =============================
+# 1. Hugging Face Authentication
+# =============================
+HF_TOKEN = os.getenv("HF_TOKEN")  # Make sure to set: export HF_TOKEN="your_token_here"
+if HF_TOKEN is None:
+    raise ValueError("⚠️ Please set your HF_TOKEN as an environment variable.")
+
+# =============================
+# 2. Load embedding + QA model
+# =============================
 embedding_model = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
-qa_model = pipeline("text-generation", model="meta-llama/Llama-3.2-3b-instruct")  # Replace with a better model if GPU is available
+qa_model = pipeline(
+    "text-generation",
+    model="meta-llama/Llama-3.2-3b-instruct",
+    token=HF_TOKEN,
+    device_map="auto"
+)
 
-# 2. Helper: extract text from files
+# =============================
+# 3. Helper: extract text from files
+# =============================
 def extract_text(file):
     text = ""
     if file.name.endswith(".pdf"):
@@ -25,72 +41,79 @@ def extract_text(file):
         text = file.read().decode("utf-8", errors="ignore")
     return text
 
-# 3. Helper: create FAISS index
+# =============================
+# 4. Helper: create FAISS index
+# =============================
 def build_faiss(text, chunk_size=500, overlap=50):
-    # Split text into chunks with overlap
     chunks = []
     for i in range(0, len(text), chunk_size - overlap):
         chunks.append(text[i:i + chunk_size])
-    
-    # Embed chunks
+
     embeddings = embedding_model.encode(chunks, convert_to_numpy=True)
-    
-    # Store in FAISS
     index = faiss.IndexFlatL2(embeddings.shape[1])
     index.add(embeddings)
-    
+
     return index, chunks
 
-# Global storage
+# =============================
+# 5. Global storage
+# =============================
 doc_index = None
 doc_chunks = None
 
-# 4. Process uploaded file
+# =============================
+# 6. Process uploaded file
+# =============================
 def upload_file(file):
     global doc_index, doc_chunks
     text = extract_text(file)
     doc_index, doc_chunks = build_faiss(text)
     return "✅ Document indexed with HyDE! You can now ask questions."
 
-# 5. HyDE RAG answering
+# =============================
+# 7. HyDE RAG answering
+# =============================
 def answer_query(query):
     global doc_index, doc_chunks
     if doc_index is None:
         return "⚠️ Please upload a document first."
-    
-    # Step 1: Generate hypothetical answer (HyDE step)
+
     hyde_prompt = f"Write a detailed, hypothetical answer to the question:\n\nQuestion: {query}\nAnswer:"
     hypo_answer = qa_model(hyde_prompt, max_length=150, num_return_sequences=1)[0]["generated_text"]
-    
-    # Step 2: Embed the hypothetical answer instead of the raw query
+
     q_emb = embedding_model.encode([hypo_answer], convert_to_numpy=True)
-    
-    # Step 3: Retrieve top 3 most relevant chunks
     D, I = doc_index.search(q_emb, k=3)
     retrieved = [doc_chunks[i] for i in I[0]]
-    
-    # Step 4: Build final prompt with context
+
     context = "\n\n".join(retrieved)
     final_prompt = f"Answer the question based on the context:\n\nContext: {context}\n\nQuestion: {query}\nAnswer:"
-    
-    # Step 5: Generate final response
+
     response = qa_model(final_prompt, max_length=200, num_return_sequences=1)[0]["generated_text"]
     return response
 
-# 6. Gradio UI
-with gr.Blocks() as demo:
-    gr.Markdown("## 📚 HyDE RAG Chatbot (Chat with Any Document)")
-    
+# =============================
+# 8. Gradio UI (Visually Appealing)
+# =============================
+with gr.Blocks(theme=gr.themes.Soft(primary_hue="indigo", secondary_hue="cyan")) as demo:
+    gr.Markdown("""
+    # 📚 HyDE RAG Chatbot
+    Talk with your documents using **Hypothetical Document Embeddings (HyDE)**.
+    Upload a PDF/DOCX/TXT and start asking questions!
+    """)
+
     with gr.Row():
-        file_input = gr.File(label="Upload Document", type="filepath")
-        upload_btn = gr.Button("Index Document")
-    
-    status = gr.Textbox(label="Status")
-    query = gr.Textbox(label="Ask a Question")
-    answer = gr.Textbox(label="Answer")
-    ask_btn = gr.Button("Get Answer")
-    
+        with gr.Column(scale=1):
+            file_input = gr.File(label="📂 Upload Document", type="filepath")
+            upload_btn = gr.Button("⚡ Index Document", variant="primary")
+            status = gr.Textbox(label="Status", interactive=False)
+
+        with gr.Column(scale=2):
+            query = gr.Textbox(label="❓ Ask a Question", placeholder="Type your question here...")
+            ask_btn = gr.Button("🚀 Get Answer", variant="primary")
+            answer = gr.Textbox(label="💡 Answer", lines=6)
+
     upload_btn.click(upload_file, inputs=file_input, outputs=status)
     ask_btn.click(answer_query, inputs=query, outputs=answer)
 
 demo.launch()
+