Update app.py

app.py

@@ -1,91 +1,25 @@
-import gradio as gr
-import os
 import faiss
-import json
-import torch
 import numpy as np
-from huggingface_hub import InferenceClient, hf_hub_download
 from sentence_transformers import SentenceTransformer
 
-# Hugging Face Credentials
-HF_REPO = "Futuresony/future_ai_12_10_2024.gguf"  # Your model repo
-FAISS_REPO = "Futuresony/future_faiss_index"  # FAISS repo
-HF_TOKEN = os.getenv('HUGGINGFACEHUB_API_TOKEN')  # Use your token
-
-# Load Chat Model
-client = InferenceClient(
-    model=HF_REPO,
-    token=HF_TOKEN
-)
-
-# Load Sentence Transformer Model for FAISS
-embedder = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
-
-# Load FAISS Index from Hugging Face
-FAISS_PATH = hf_hub_download(repo_id=FAISS_REPO, filename="asa_faiss.index", repo_type="model", token=HF_TOKEN)
+# 🔹 Load FAISS Index
+FAISS_PATH = "asa_faiss.index"
 faiss_index = faiss.read_index(FAISS_PATH)
 
-# Load FAISS Text Data
-TEXT_DATA_PATH = hf_hub_download(repo_id=FAISS_REPO, filename="asa_text_data.npy", repo_type="model", token=HF_TOKEN)
-text_data = np.load(TEXT_DATA_PATH, allow_pickle=True)
-
-def retrieve_faiss_knowledge(user_query, top_k=3):
-    """Retrieve the most relevant FAISS knowledge based on user input."""
-    query_embedding = embedder.encode([user_query], convert_to_tensor=True).cpu().numpy()
-    distances, indices = faiss_index.search(query_embedding, top_k)
-
-    retrieved_texts = []
-    print("\n🔍 DEBUG: FAISS Retrieved Indices and Distances")
-    print(indices, distances)  # 🔥 Check if FAISS is retrieving valid results
-
-    for idx in indices[0]:  # Extract top_k results
-        if idx != -1:  # Ensure valid index
-            retrieved_texts.append(text_data[idx])  # ✅ Retrieve actual stored FAISS text!
-
-    return "\n".join(retrieved_texts) if retrieved_texts else "**No relevant FAISS data found.**"
-
-def format_alpaca_prompt(user_input, system_prompt, history, faiss_knowledge=""):
-    """Formats input in Alpaca/LLaMA style"""
-    history_str = "\n".join([f"### Instruction:\n{h[0]}\n### Response:\n{h[1]}" for h in history])
-    faiss_context = f"\n### Retrieved Knowledge:\n{faiss_knowledge}" if faiss_knowledge else ""
-
-    prompt = f"""{system_prompt}
-{history_str}
-
-### Instruction:
-{user_input}
-{faiss_context}
-
-### Response:
-"""
-    return prompt
-
-def respond(message, history, system_message, max_tokens, temperature, top_p):
-    faiss_knowledge = retrieve_faiss_knowledge(message, top_k=3)  # ✅ Get FAISS data
-    formatted_prompt = format_alpaca_prompt(message, system_message, history, faiss_knowledge)
+# 🔹 Load Embedding Model
+embedder = SentenceTransformer("sentence-transformers/all-MiniLM-L6-v2")
 
-    response = client.text_generation(
-        formatted_prompt,
-        max_new_tokens=max_tokens,
-        temperature=temperature,
-        top_p=top_p,
-    )
+# 🔹 Test Query
+test_query = "Where is ASA Microfinance located?"
+query_embedding = embedder.encode([test_query], convert_to_tensor=True).cpu().numpy()
 
-    cleaned_response = response.split("### Response:")[-1].strip()
-    
-    history.append((message, cleaned_response))  # ✅ Update history with new response
-    
-    yield cleaned_response  # ✅ Return chatbot's answer
+# 🔹 Search FAISS
+distances, indices = faiss_index.search(query_embedding, 3)  # Retrieve top 3 matches
 
-demo = gr.ChatInterface(
-    respond,
-    additional_inputs=[
-        gr.Textbox(value="You are a friendly Chatbot.", 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)"),
-    ],
-)
+# 🔹 Print Results
+print("🔹 FAISS Search Results:")
+for idx in indices[0]:
+    print(f"Index: {idx}")
 
-if __name__ == "__main__":
-    demo.launch()
+print("🔹 FAISS Distances:")
+print(distances)