Update app.py

app.py

@@ -1,72 +1,56 @@
 import gradio as gr
-from transformers import AutoModelForCausalLM, AutoTokenizer
 import torch
+from transformers import AutoModelForCausalLM, AutoTokenizer
 
-# =========================
-# MODEL LOAD
-# =========================
-MODEL_NAME = "Qwen/Qwen1.5-1.8B-Chat"  # ya jo tum use kar rahe ho
+# 1. Lightweight Model (फ्री CPU सर्वर के लिए एकदम परफेक्ट)
+REPO_ID = "Qwen/Qwen2.5-0.5B-Instruct"
 
-print("[System]: Loading model...")
+print("लोड हो रहा है... कृपया प्रतीक्षा करें...")
+tokenizer = AutoTokenizer.from_pretrained(REPO_ID)
 
-tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME)
+# torch.bfloat16 मेमोरी को आधा कर देता है जिससे क्रैश नहीं होता
 model = AutoModelForCausalLM.from_pretrained(
-    MODEL_NAME,
-    torch_dtype=torch.float16 if torch.cuda.is_available() else torch.float32,
-    device_map="auto"
+    REPO_ID,
+    torch_dtype=torch.bfloat16, 
+    device_map="cpu",
+    low_cpu_mem_usage=True
 )
-
-print("[System]: Model loaded!")
-
-# =========================
-# CHAT FUNCTION
-# =========================
-def chat(user_message, history):
-    if history is None:
-        history = []
-
-    # Convert history to text format
-    conversation = ""
-    for user, bot in history:
-        conversation += f"User: {user}\nAssistant: {bot}\n"
-
-    conversation += f"User: {user_message}\nAssistant:"
-
-    inputs = tokenizer(conversation, return_tensors="pt").to(model.device)
-
+print("मॉडल सफलतापूर्वक लोड हो गया!")
+
+# 2. Simple Core Logic (सिर्फ रिस्पॉन्स जनरेट करने के लिए)
+def generate_response(message, history):
+    # Gradio की हिस्ट्री को Qwen के फॉर्मेट में बदलना
+    messages = []
+    for user_msg, bot_msg in history:
+        messages.append({"role": "user", "content": user_msg})
+        messages.append({"role": "assistant", "content": bot_msg})
+    
+    # नया मैसेज जोड़ना
+    messages.append({"role": "user", "content": message})
+
+    text = tokenizer.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
+    inputs = tokenizer([text], return_tensors="pt")
+
+    # Fast Inference सेटिंग्स
     outputs = model.generate(
         **inputs,
-        max_new_tokens=200,
+        max_new_tokens=150, # जवाब की लिमिट ताकि सर्वर न अटके
         temperature=0.7,
-        top_p=0.9,
         do_sample=True
     )
+    
+    response_ids = outputs[0][inputs.input_ids.shape[1]:]
+    response = tokenizer.decode(response_ids, skip_special_tokens=True)
+    
+    return response
+
+# 3. Minimal UI & API Setup (ChatInterface सब खुद हैंडल करेगा)
+demo = gr.ChatInterface(
+    fn=generate_response,
+    title="⚡ Qwen 0.5B Fast API",
+    description="Minimal, Fast & Stable Backend for API Calls",
+)
 
-    response = tokenizer.decode(outputs[0], skip_special_tokens=True)
-
-    # Extract only latest answer
-    response = response.split("Assistant:")[-1].strip()
-
-    history.append((user_message, response))
-
-    return history, history
-
-
-# =========================
-# UI
-# =========================
-with gr.Blocks() as demo:
-    gr.Markdown("# 🤖 Qwen Chatbot")
-
-    chatbot = gr.Chatbot()  # ✅ FIXED (no type="messages")
-
-    msg = gr.Textbox(placeholder="Type your message here...")
-    clear = gr.Button("Clear")
-
-    msg.submit(chat, [msg, chatbot], [chatbot, chatbot])
-    clear.click(lambda: None, None, chatbot, queue=False)
-
-# =========================
-# LAUNCH
-# =========================
-demo.launch()
+# show_error=True से हमें असली एरर दिखेगा
+if __name__ == "__main__":
+    demo.launch(show_error=True)