Update app.py

app.py

@@ -3,9 +3,8 @@ import faiss
 import pickle
 import numpy as np
 from sentence_transformers import SentenceTransformer
-from transformers import AutoTokenizer, AutoModelForCausalLM, pipeline
-
 import os
+
 print("Files in current directory:", os.listdir())
 
 # -----------------------------
@@ -17,34 +16,11 @@ index = faiss.read_index("faiss_index.bin")
 chunks = pickle.load(open("chunks.pkl", "rb"))
 metadata = pickle.load(open("metadata.pkl", "rb"))
 
-# -----------------------------
-# Load free HF small LLM
-# -----------------------------
-# Using distilgpt2 as it doesn't need a token
-model_name = "distilgpt2"
-
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = AutoModelForCausalLM.from_pretrained(
-    model_name,
-)
-
-generator = pipeline(
-    "text-generation",
-    model=model,
-    tokenizer=tokenizer,
-    max_new_tokens=150,
-    do_sample=True,
-    temperature=0.6
-)
-
-print("LLM loaded successfully!")
-
 # -----------------------------
 # Intent detection
 # -----------------------------
 def detect_query(query):
     query = query.lower()
-
     animal = None
     topic = None
 
@@ -78,7 +54,6 @@ def retrieve_context(query):
         filtered_indices = list(range(len(chunks)))
 
     query_embedding = embed_model.encode([query])
-
     filtered_embeddings = np.array([index.reconstruct(i) for i in filtered_indices])
     distances = np.linalg.norm(filtered_embeddings - query_embedding, axis=1)
     top_indices = distances.argsort()[:2]
@@ -88,47 +63,16 @@ def retrieve_context(query):
         real_index = filtered_indices[idx]
         context += chunks[real_index] + "\n"
 
-    return context
+    return context.strip()
 
 # -----------------------------
-# Chat function
+# Chat function (RAG only)
 # -----------------------------
 def chat(user_input):
     context = retrieve_context(user_input)
-
-    prompt = f"""
-You are a livestock expert assistant.
-
-Use ONLY the information below to answer.
-If answer is not present, say "I don't know".
-
-Context:
-{context}
-
-Question:
-{user_input}
-
-Answer in short, clear, and complete sentences.
-"""
-
-    response = generator(
-        prompt,
-        max_new_tokens=100,
-        do_sample=True,
-        temperature=0.6,
-        pad_token_id=tokenizer.eos_token_id
-    )
-
-    text = response[0]["generated_text"]
-
-    # Remove prompt repetition
-    if prompt.strip() in text:
-        text = text.split(prompt.strip())[-1].strip()
-
-    # Keep only first paragraph or sentence to avoid repetition
-    text = text.split("\n")[0].strip()
-
-    return text
+    if not context:
+        return "I don't know."
+    return context
 
 # -----------------------------
 # Gradio UI
@@ -137,5 +81,6 @@ gr.Interface(
     fn=chat,
     inputs="text",
     outputs="text",
-    title="Livestock Chatbot"
+    title="Livestock Chatbot (RAG only)",
+    description="This chatbot answers livestock questions using only the retrieved data. No AI model is used."
 ).launch()