Update app.py

app.py

@@ -8,16 +8,16 @@ from chromadb.config import Settings
 from transformers import pipeline
 
 # Device setup
-device = -1  # Force CPU use
+device = -1  # Use CPU
 print("Device set to: CPU")
 
 # Load CSV data
-df = pd.read_csv("iec_college_data.csv").dropna(subset=["content"]).reset_index(drop=True)
+df = pd.read_csv("/mnt/data/iec_college_data.csv").dropna(subset=["content"]).reset_index(drop=True)
 
 # Load embedding model on CPU
 embed_model = SentenceTransformer("all-MiniLM-L6-v2", device="cpu")
 
-# ChromaDB - use updated client format
+# ChromaDB setup
 chroma_client = chromadb.PersistentClient(path="./chroma_db")
 collection_name = "iec_data"
 
@@ -45,8 +45,8 @@ if collection.count() == 0:
             print(f"Indexed {idx}/{len(df)}")
     print("Indexing complete.")
 
-# QA model: Use lighter model on CPU
-qa_pipeline = pipeline("text2text-generation", model="google/flan-t5-small", device=device)
+# Use lightweight extractive QA model
+qa_pipeline = pipeline("question-answering", model="distilbert-base-uncased", device=device)
 
 # QA function
 def answer_question(user_question):
@@ -55,11 +55,10 @@ def answer_question(user_question):
     context = "\n".join(results["documents"][0])
     if len(context.split()) > 400:
         context = " ".join(context.split()[:400])
-    prompt = f"You are an assistant for IEC College. Use the info below.\n\nContext:\n{context}\n\nQuestion: {user_question}\nAnswer:"
-    result = qa_pipeline(prompt, max_new_tokens=200)[0]["generated_text"]
-    return result.strip()
+    result = qa_pipeline(question=user_question, context=context)
+    return result["answer"]
 
-# Gradio interface
+# Gradio UI
 iface = gr.Interface(
     fn=answer_question,
     inputs=gr.Textbox(lines=2, placeholder="Ask about IEC College..."),