Update app.py

app.py

@@ -2,16 +2,15 @@ import gradio as gr
 from transformers import pipeline
 from datasets import load_dataset
 
-# Choose your model here – change if needed
-model_name = "Qwen/Qwen1.5-7B-Chat"
+# ✅ Use smaller, memory-friendly model
+model_name = "Qwen/Qwen2.5-3B"
 
-# Load the text generation pipeline
+# Load model pipeline
 pipe = pipeline("text-generation", model=model_name, device=0)
 
-# Load GTA dataset (optional)
+# Load GTA benchmark (optional)
 gta_data = load_dataset("open-compass/GTA", split="test")
 
-# Inference function
 def run_model(input_text, use_gta_index):
     if use_gta_index:
         try:
@@ -27,8 +26,8 @@ def run_model(input_text, use_gta_index):
 
 # Gradio UI
 with gr.Blocks() as demo:
-    gr.Markdown("# 🤖 GTA-style Reasoning with Hugging Face Models")
-    gr.Markdown("Enter a question or choose a sample from the GTA benchmark.")
+    gr.Markdown("# 🤖 GTA-style Reasoning with Qwen 2.5-3B")
+    gr.Markdown("Test queries from the GTA benchmark or enter your own.")
     with gr.Row():
         input_text = gr.Textbox(label="Enter a question or GTA index (e.g., 0–228)")
         use_sample = gr.Checkbox(label="Use as GTA index", value=False)
@@ -37,4 +36,4 @@ with gr.Blocks() as demo:
     
     run_button.click(run_model, inputs=[input_text, use_sample], outputs=output_text)
 
-demo.launch()
+demo.launch()