Update app.py

app.py

@@ -1,28 +1,39 @@
-
 # =========================================
-# RAG QnA System (ACCURATE - BERT QA)
+# RAG QnA System (FIXED FOR HF SPACES)
 # =========================================
 
-
 import gradio as gr
 import numpy as np
 import faiss
+import os
 
 from sentence_transformers import SentenceTransformer
-from transformers import pipeline
+from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
+import torch
 
 # -----------------------------
-# 1. Load Documents
+# 1. Load Documents (FIXED)
 # -----------------------------
 def load_documents(file_path):
-    with open(file_path, "r", encoding="utf-8") as f:
-        text = f.read()
+    if not os.path.exists(file_path):
+        return ["No document found."]
+    
+    try:
+        with open(file_path, "r", encoding="utf-8") as f:
+            text = f.read()
+    except:
+        with open(file_path, "r", encoding="latin-1") as f:
+            text = f.read()
+
     return text.split("\n\n")
 
 
 def chunk_text(text, chunk_size=120):
     words = text.split()
-    return [" ".join(words[i:i+chunk_size]) for i in range(0, len(words), chunk_size)]
+    return [
+        " ".join(words[i:i+chunk_size])
+        for i in range(0, len(words), chunk_size)
+    ]
 
 
 documents = load_documents("data/data.txt")
@@ -43,15 +54,15 @@ index = faiss.IndexFlatL2(dimension)
 index.add(np.array(embeddings))
 
 # -----------------------------
-# 3. Load QA Model (BERT)
+# 3. GENERATIVE MODEL (FIXED)
 # -----------------------------
-qa_pipeline = pipeline(
-    "question-answering",
-    model="distilbert-base-cased-distilled-squad"
-)
+model_name = "google/flan-t5-base"
+
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
 
 # -----------------------------
-# 4. RAG Function
+# 4. RAG Function (FIXED)
 # -----------------------------
 def rag_query(query):
     if not query.strip():
@@ -64,19 +75,30 @@ def rag_query(query):
     retrieved_docs = [all_chunks[i] for i in I[0]]
     context = " ".join(retrieved_docs)
 
-    # Extract exact answer
-    result = qa_pipeline(
-        question=query,
-        context=context
-    )
+    # Prompt for model
+    prompt = f"""
+    Answer the question ONLY using the context below.
+    If the answer is not present, say "Not found in document".
 
-    answer = result["answer"]
+    Context:
+    {context}
 
-    # If low confidence → avoid hallucination
-    if result["score"] < 0.2:
-        answer = "Not found in document"
+    Question:
+    {query}
+    """
 
-    return f"Answer:\n{answer}\n\nConfidence: {result['score']:.2f}\n\nContext:\n{context}"
+    inputs = tokenizer(prompt, return_tensors="pt", truncation=True)
+
+    outputs = model.generate(
+        **inputs,
+        max_new_tokens=150,
+        do_sample=True,
+        temperature=0.7
+    )
+
+    answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
+
+    return f"Answer:\n{answer}\n\nContext:\n{context}"
 
 # -----------------------------
 # 5. Gradio UI
@@ -85,208 +107,12 @@ iface = gr.Interface(
     fn=rag_query,
     inputs=gr.Textbox(lines=2, placeholder="Ask your question..."),
     outputs="text",
-    title="📚 RAG QnA System (Accurate)",
-    description="Retriever + BERT QA → No hallucination"
+    title="📚 RAG QnA System (Fixed)",
+    description="Retriever + FLAN-T5 (Works on Hugging Face Spaces)"
 )
 
 # -----------------------------
 # 6. Launch
 # -----------------------------
 if __name__ == "__main__":
-    iface.launch()
-
-
-# # =========================================
-# # RAG QnA System (ACCURATE - BERT QA)
-# # =========================================
-
-# import gradio as gr
-# import numpy as np
-# import faiss
-
-# from sentence_transformers import SentenceTransformer
-# from transformers import pipeline
-
-# # -----------------------------
-# # 1. Load Documents
-# # -----------------------------
-# def load_documents(file_path):
-#     with open(file_path, "r", encoding="utf-8") as f:
-#         text = f.read()
-#     return text.split("\n\n")
-
-
-# def chunk_text(text, chunk_size=120):
-#     words = text.split()
-#     return [" ".join(words[i:i+chunk_size]) for i in range(0, len(words), chunk_size)]
-
-
-# documents = load_documents("data/data.txt")
-
-# all_chunks = []
-# for doc in documents:
-#     all_chunks.extend(chunk_text(doc))
-
-# # -----------------------------
-# # 2. Embeddings + FAISS
-# # -----------------------------
-# embedder = SentenceTransformer("all-MiniLM-L6-v2")
-
-# embeddings = embedder.encode(all_chunks)
-
-# dimension = embeddings.shape[1]
-# index = faiss.IndexFlatL2(dimension)
-# index.add(np.array(embeddings))
-
-# # -----------------------------
-# # 3. Load QA Model (BERT)
-# # -----------------------------
-# qa_pipeline = pipeline(
-#     "question-answering",
-#     model="distilbert-base-cased-distilled-squad"
-# )
-
-# # -----------------------------
-# # 4. RAG Function
-# # -----------------------------
-# def rag_query(query):
-#     if not query.strip():
-#         return "Please enter a question."
-
-#     # Retrieve relevant chunks
-#     query_embedding = embedder.encode([query])
-#     D, I = index.search(np.array(query_embedding), k=5)
-
-#     retrieved_docs = [all_chunks[i] for i in I[0]]
-#     context = " ".join(retrieved_docs)
-
-#     # Extract exact answer
-#     result = qa_pipeline(
-#         question=query,
-#         context=context
-#     )
-
-#     answer = result["answer"]
-
-#     # If low confidence → avoid hallucination
-#     if result["score"] < 0.2:
-#         answer = "Not found in document"
-
-#     return f"Answer:\n{answer}\n\nConfidence: {result['score']:.2f}\n\nContext:\n{context}"
-
-# # -----------------------------
-# # 5. Gradio UI
-# # -----------------------------
-# iface = gr.Interface(
-#     fn=rag_query,
-#     inputs=gr.Textbox(lines=2, placeholder="Ask your question..."),
-#     outputs="text",
-#     title="📚 RAG QnA System (Accurate)",
-#     description="Retriever + BERT QA → No hallucination"
-# )
-
-# # -----------------------------
-# # 6. Launch
-# # -----------------------------
-# if __name__ == "__main__":
-#     iface.launch()
- 
-
-# #import gradio as gr
-# # import numpy as np
-# # import faiss
-
-# # from sentence_transformers import SentenceTransformer
-# # from transformers import AutoTokenizer, AutoModelForSeq2SeqLM
-# # import torch
-
-# # # -----------------------------
-# # # 1. Load Documents
-# # # -----------------------------
-# # def load_documents(file_path):
-# #     with open(file_path, "r", encoding="utf-8") as f:
-# #         text = f.read()
-# #     return text.split("\n\n")
-
-
-# # def chunk_text(text, chunk_size=100):
-# #     words = text.split()
-# #     return [" ".join(words[i:i+chunk_size]) for i in range(0, len(words), chunk_size)]
-
-
-# # documents = load_documents("data/data.txt")
-
-# # all_chunks = []
-# # for doc in documents:
-# #     all_chunks.extend(chunk_text(doc))
-
-# # # -----------------------------
-# # # 2. Embeddings + FAISS
-# # # -----------------------------
-# # embedder = SentenceTransformer("all-MiniLM-L6-v2")
-
-# # embeddings = embedder.encode(all_chunks)
-
-# # dimension = embeddings.shape[1]
-# # index = faiss.IndexFlatL2(dimension)
-# # index.add(np.array(embeddings))
-
-# # # -----------------------------
-# # # 3. Load Model (FIXED)
-# # # -----------------------------
-# # model_name = "google/flan-t5-small"
-
-# # tokenizer = AutoTokenizer.from_pretrained(model_name)
-# # model = AutoModelForSeq2SeqLM.from_pretrained(model_name)
-
-# # # -----------------------------
-# # # 4. RAG Function
-# # # -----------------------------
-# # def rag_query(query):
-# #     if not query.strip():
-# #         return "Please enter a question."
-
-# #     query_embedding = embedder.encode([query])
-
-# #     D, I = index.search(np.array(query_embedding), k=3)
-# #     retrieved_docs = [all_chunks[i] for i in I[0]]
-
-# #     context = " ".join(retrieved_docs)
-
-# #     input_text = f"""
-# #     Answer the question based only on the context below.
-
-# #     Context: {context}
-
-# #     Question: {query}
-# #     """
-
-# #     inputs = tokenizer(input_text, return_tensors="pt", truncation=True)
-
-# #     outputs = model.generate(
-# #         **inputs,
-# #         max_length=120,
-# #         do_sample=True,
-# #         temperature=0.7
-# #     )
-
-# #     answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
-
-# #     return f"Answer:\n{answer}\n\nContext:\n{context}"
-
-# # # -----------------------------
-# # # 5. Gradio UI
-# # # -----------------------------
-# # iface = gr.Interface(
-# #     fn=rag_query,
-# #     inputs=gr.Textbox(lines=2, placeholder="Ask your question..."),
-# #     outputs="text",
-# #     title="📚 RAG QnA System",
-# #     description="Ask questions based on your document"
-# # )
-
-# # # -----------------------------
-# # # 6. Launch
-# # # -----------------------------
-# # if __name__ == "__main__":
-# #     iface.launch()
+    iface.launch()