Update app.py

app.py

@@ -1,5 +1,4 @@
 import streamlit as st
-import numpy as np
 import torch
 from transformers import DistilBertTokenizer, DistilBertForQuestionAnswering
 
@@ -11,35 +10,63 @@ def load_model():
     tokenizer = DistilBertTokenizer.from_pretrained("distilbert-base-uncased-distilled-squad")
     return model, tokenizer
 
+def generate_prompt(question, text):
+    """Enhance the input prompt to guide the model better."""
+    return (
+        f"Context: {text}\n\n"
+        f"Instruction: Read the above context carefully and extract the most relevant answer.\n"
+        f"Question: {question}\n"
+        f"Answer:"
+    )
+
 def get_answer(question, text, tokenizer, model):
-    if "your name" or "who are you" or "about you" in question.lower():
+    # Special case for bot identity
+    if any(phrase in question.lower() for phrase in ["your name", "who are you", "about you"]):
         return "I am Numini, NativUttarMini, created by Sanju Debnath at University of Calcutta."
-    
-    inputs = tokenizer(question, text, return_tensors="pt", truncation=True, padding=True)
+
+    # Enhance the input for better response
+    prompt_text = generate_prompt(question, text)
+
+    # Tokenize with truncation for better handling of large text
+    inputs = tokenizer(prompt_text, return_tensors="pt", truncation=True, padding=True, max_length=512)
+
     with torch.no_grad():
         outputs = model(**inputs)
+
     start = torch.argmax(outputs.start_logits)
     end = torch.argmax(outputs.end_logits) + 1
-    ans_tokens = inputs.input_ids[0][start:end]
-    answer = tokenizer.decode(ans_tokens, skip_special_tokens=True)
+
+    # Handle cases where model is uncertain
+    if start >= end:
+        return "I couldn't find a clear answer in the given text."
+
+    answer = tokenizer.decode(inputs.input_ids[0][start:end], skip_special_tokens=True)
+
+    # Post-process answer for better readability
+    if len(answer.split()) < 3:  # Model sometimes returns incomplete answers
+        return "I'm not sure about the exact answer. Can you try rephrasing the question?"
+    
     return answer
 
 def main():
-    st.write("# Question Answering Tool \n"
-             "This tool will help you find answers to your questions about the text you provide. \n"
-             "Please enter your question and the text you want to search in the boxes below.")
+    st.title("🔎 Advanced Question Answering Tool")
+    st.write("Ask a question based on the given text, and I'll extract the best possible answer.")
+
     model, tokenizer = load_model()
 
     with st.form("qa_form"):
-        text = st.text_area("Enter your text here")
-        question = st.text_input("Enter your question here")
-        
-        if st.form_submit_button("Submit"):
-            data_load_state = st.text('Let me think about that...')
-            answer = get_answer(question, text, tokenizer, model)
-            if answer.strip() == "":
-                data_load_state.text("Sorry but I don't know the answer to that question")
+        text = st.text_area("📜 Enter the text/document:", height=200)
+        question = st.text_input("❓ Enter your question:")
+
+        if st.form_submit_button("🔍 Get Answer"):
+            if not text.strip():
+                st.warning("⚠️ Please enter some text to analyze.")
+            elif not question.strip():
+                st.warning("⚠️ Please enter a question.")
             else:
-                data_load_state.text(answer)
+                st.text("🤖 Thinking...")
+                answer = get_answer(question, text, tokenizer, model)
+                st.success(f"✅ Answer: {answer}")
 
 main()
+