Update app.py

app.py

@@ -1,18 +1,23 @@
 import gradio as gr
 from transformers import pipeline
 
-# Setting up the Coretex brain
-# Using Qwen 2.5 3B - very powerful and free to use immediately
+# Setting up Coretex - Specifically for CPU Free Tier
 model_id = "Qwen/Qwen2.5-3B-Instruct"
-pipe = pipeline("text-generation", model=model_id, device_map="auto")
+
+# We removed device_map="auto" to fix the error and added a simpler loader
+pipe = pipeline(
+    "text-generation", 
+    model=model_id, 
+    model_kwargs={"low_cpu_mem_usage": True}
+)
 
 def coretex_chat(user_input):
     messages = [{"role": "user", "content": user_input}]
-    # Coretex thinks here...
-    response = pipe(messages, max_new_tokens=256)
+    # Coretex is thinking... 
+    # Note: On CPU, this might take 10-20 seconds per answer.
+    response = pipe(messages, max_new_tokens=128)
     return response[0]['generated_text'][-1]['content']
 
-# The interface humans see (we'll add the API for your bot later)
 demo = gr.Interface(
     fn=coretex_chat, 
     inputs=gr.Textbox(label="Message Coretex"),