Update app.py

app.py

@@ -1,83 +1,14 @@
-import gradio as gr
-import os
-import json
 import faiss
 import numpy as np
-import torch
-from sentence_transformers import SentenceTransformer
-from huggingface_hub import InferenceClient, hf_hub_download
 
-# 🔹 Hugging Face Credentials
-HF_REPO = "Futuresony/future_ai_12_10_2024.gguf"
-HF_TOKEN = os.getenv('HUGGINGFACEHUB_API_TOKEN')  # Ensure this is set in your environment
-
-# 🔹 FAISS Index Path
+# Load FAISS index
 FAISS_PATH = "asa_faiss.index"
+index = faiss.read_index(FAISS_PATH)
 
-# 🔹 Load Sentence Transformer for Embeddings
-embedder = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
-
-# 🔹 Load FAISS Index from Hugging Face
-faiss_local_path = hf_hub_download(HF_REPO, "asa_faiss.index", token=HF_TOKEN)
-faiss_index = faiss.read_index(faiss_local_path)
-
-# 🔹 Initialize Hugging Face Model Client
-client = InferenceClient(model=HF_REPO, token=HF_TOKEN)
-
-# 🔹 Retrieve Relevant FAISS Data
-def retrieve_faiss_knowledge(user_query, top_k=3):
-    query_embedding = embedder.encode([user_query], convert_to_tensor=True).cpu().numpy()
-    distances, indices = faiss_index.search(query_embedding, top_k)
-
-    retrieved_texts = []
-    for idx in indices[0]:  # Extract top_k results
-        if idx != -1:  # Ensure valid index
-            retrieved_texts.append(f"Example {idx}: (Extracted FAISS Data)")
-
-    return "\n".join(retrieved_texts) if retrieved_texts else "**No relevant FAISS data found.**"
-
-# 🔹 Chatbot Response Function (Forcing FAISS Context)
-def respond(message, history, system_message, max_tokens, temperature, top_p):
-    faiss_context = retrieve_faiss_knowledge(message)
-
-    # 🔥 Force the model to use FAISS
-    full_prompt = f"""### System Instruction:
-You MUST use the provided FAISS data to generate your response.
-If no FAISS data is found, return "I don't have enough information."
-
-### Retrieved FAISS Data:
-{faiss_context}
-
-### User Query:
-{message}
-
-### Response:
-"""
-
-    response = client.text_generation(
-        full_prompt,
-        max_new_tokens=max_tokens,
-        temperature=temperature,
-        top_p=top_p,
-    )
-
-    # ✅ Extract only the model-generated response
-    cleaned_response = response.split("### Response:")[-1].strip()
-
-    history.append((message, cleaned_response))  # ✅ Update chat history
-
-    yield cleaned_response  # ✅ Output the response
+# Example query vector (random, replace with actual embedding from your model)
+query_vector = np.random.rand(1, index.d).astype('float32')
 
-# 🔹 Gradio Chat Interface
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a knowledge assistant that must use FAISS context.", label="System message"),
-        gr.Slider(minimum=1, maximum=250, value=128, step=1, label="Max new tokens"),
-        gr.Slider(minimum=0.1, maximum=4.0, value=0.9, step=0.1, label="Temperature"),
-        gr.Slider(minimum=0.1, maximum=1.0, value=0.99, step=0.01, label="Top-p (nucleus sampling)"),
-    ],
-)
+# Search FAISS index
+D, I = index.search(query_vector, k=1)  # k=1 means get 1 nearest neighbor
 
-if __name__ == "__main__":
-    demo.launch()
+print(f"Closest match index: {I[0][0]}, Distance: {D[0][0]}")