Update app.py

app.py

@@ -1,81 +1,47 @@
-import os
-import faiss
-import numpy as np
-import torch
 import gradio as gr
-from transformers import AutoModelForCausalLM, AutoTokenizer
-
-# ✅ Fix: Use AutoTokenizer instead of GemmaTokenizer
-tokenizer = AutoTokenizer.from_pretrained(HF_REPO, token=HF_TOKEN)
-
-from huggingface_hub import hf_hub_download
-
-# 🔹 Hugging Face Credentials
-HF_REPO = "Futuresony/my_model"  # Ensure this is correct
-HF_TOKEN = os.getenv('HUGGINGFACEHUB_API_TOKEN')  # Ensure this is set in your environment
-
-# 🔹 FAISS Index Path
-FAISS_PATH = "asa_faiss.index"
-DATASET_PATH = "responses.txt"  # Ensure this file contains indexed responses
-
-# ✅ Load FAISS Index from Hugging Face if not available locally
-if not os.path.exists(FAISS_PATH):
-    print("🔄 Downloading FAISS index...")
-    FAISS_PATH = hf_hub_download(HF_REPO, "asa_faiss.index", token=HF_TOKEN)
-
-print(f"📂 Loading FAISS index from {FAISS_PATH}...")
-faiss_index = faiss.read_index(FAISS_PATH)
-print("✅ FAISS index loaded successfully!")
-
-# ✅ Load responses dataset
-if os.path.exists(DATASET_PATH):
-    with open(DATASET_PATH, "r", encoding="utf-8") as f:
-        dataset = f.readlines()
-    print("✅ Responses dataset loaded!")
-else:
-    print(f"⚠️ Warning: {DATASET_PATH} not found!")
-    dataset = []
-
-# ✅ Load model & tokenizer
-print("🔄 Loading tokenizer and model...")
-tokenizer = GemmaTokenizer.from_pretrained(HF_REPO, token=HF_TOKEN)
-model = AutoModelForCausalLM.from_pretrained(HF_REPO, token=HF_TOKEN)
-print("✅ Model and tokenizer loaded!")
-
-# 🔹 Set FAISS distance threshold (lower values = more strict matches)
-THRESHOLD = 80  # Adjusted threshold for better accuracy
-
-def embed(text):
-    """Convert text to FAISS-compatible vector."""
-    tokens = tokenizer.encode(text, add_special_tokens=True)
-    return np.array(tokens, dtype=np.float32).reshape(1, -1)
-
-def chatbot_response(user_query):
-    """Fetches response from FAISS or falls back to the model."""
-    query_vector = embed(user_query)
-    D, I = faiss_index.search(query_vector, k=1)
-
-    print(f"🔍 Closest FAISS match index: {I[0][0]}, Distance: {D[0][0]}")
-
-    if D[0][0] < THRESHOLD and 0 <= I[0][0] < len(dataset):
-        response = dataset[I[0][0]].strip()
-        print("✅ FAISS response used!")
-    else:
-        print("⚠️ FAISS match too weak, generating response using model.")
-        inputs = tokenizer(user_query, return_tensors="pt")
-        outputs = model.generate(**inputs, max_new_tokens=150)
-        response = tokenizer.decode(outputs[0], skip_special_tokens=True)
-
-    return response
-
-# 🔹 Gradio UI
-iface = gr.Interface(
-    fn=chatbot_response,
-    inputs="text",
-    outputs="text",
-    title="ASA Microfinance Chatbot",
-    description="A chatbot that provides information using FAISS and a language model."
+from huggingface_hub import InferenceClient
+
+client = InferenceClient("Futuresony/future_ai_12_10_2024.gguf")
+
+def format_alpaca_prompt(user_input, system_prompt, history):
+    """Formats input in Alpaca/LLaMA style"""
+    history_str = "\n".join([f"### Instruction:\n{h[0]}\n### Response:\n{h[1]}" for h in history])
+    prompt = f"""{system_prompt}
+{history_str}
+
+### Instruction:
+{user_input}
+
+### Response:
+"""
+    return prompt
+
+def respond(message, history, system_message, max_tokens, temperature, top_p):
+    formatted_prompt = format_alpaca_prompt(message, system_message, history)
+
+    response = client.text_generation(
+        formatted_prompt,
+        max_new_tokens=max_tokens,
+        temperature=temperature,
+        top_p=top_p,
+    )
+
+    # ✅ Extract only the response
+    cleaned_response = response.split("### Response:")[-1].strip()
+    
+    history.append((message, cleaned_response))  # ✅ Update history with the new message and response
+    
+    yield cleaned_response  # ✅ Output only the answer
+
+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)"),
+    ],
 )
 
 if __name__ == "__main__":
-    iface.launch()
+    demo.launch()