Update app.py

app.py

@@ -1,17 +1,36 @@
-import gradio as gr
-from transformers import pipeline
+from flask import Flask, request, jsonify, render_template
+from flask_cors import CORS
+from llama_cpp import Llama
+import os
 
-# Load chatbot pipeline từ Hugging Face
-chatbot = pipeline("text-generation", model="TinyLlama/TinyLlama-1.1B-Chat-v1.0")
+app = Flask(__name__)
+CORS(app)
 
-# Hàm xử lý chat
-def chat_with_ai(user_message):
-    prompt = f"User: {user_message}\nAssistant:"
-    response = chatbot(prompt, max_length=200, do_sample=True, temperature=0.7)
-    return response[0]["generated_text"].split("Assistant:")[-1].strip()
+# Đường dẫn tới model GGUF
+MODEL_PATH = "models/mistral-7b-instruct-v0.2.Q4_K_M.gguf"
 
-# Giao diện Gradio
-iface = gr.Interface(fn=chat_with_ai, inputs="text", outputs="text", title="TinyLlama Chatbot")
+# Kiểm tra model có tồn tại không
+if not os.path.exists(MODEL_PATH):
+    raise FileNotFoundError(f"Không tìm thấy model tại {MODEL_PATH}")
 
-# Chạy ứng dụng
-iface.launch()
+# Load model GGUF
+llm = Llama(model_path=MODEL_PATH, n_ctx=4096, n_threads=8, n_batch=512)
+
+# Xử lý chat
+def get_answer(user_text):
+    response = llm(f"User: {user_text}\nAssistant:", max_tokens=150, temperature=0.7)
+    return response["choices"][0]["text"]
+
+@app.route('/')
+def index():
+    return render_template('index.html')
+
+@app.route('/chat', methods=['POST'])
+def chat():
+    data = request.json
+    user_message = data.get('message', '')
+    bot_reply = get_answer(user_message)
+    return jsonify({'FROM': 'AI Assistant', 'MESSAGE': bot_reply})
+
+if __name__ == '__main__':
+    app.run(host="0.0.0.0", port=7860, debug=True)