Update app.py

app.py

@@ -7,12 +7,15 @@ model_name = "PierreJousselin/gpt2"  # Replace with the name you used on Hugging
 tokenizer = AutoTokenizer.from_pretrained(model_name)
 model = AutoModelForCausalLM.from_pretrained(model_name, device_map="cpu")  # Force model to load on CPU
 
+# Ensure pad_token_id is set to eos_token_id to avoid errors
+model.config.pad_token_id = model.config.eos_token_id
+
 # Function for generating responses using the model
 def generate_response(prompt):
     # Tokenize input prompt
     inputs = tokenizer(prompt, return_tensors="pt", padding=True, truncation=True, max_length=128)
     
-    # Make sure the inputs are moved to the CPU
+    # Ensure the inputs are moved to the CPU
     input_ids = inputs["input_ids"].to("cpu")
     
     # Generate output (ensure it's on CPU)
@@ -22,13 +25,14 @@ def generate_response(prompt):
     response = tokenizer.decode(output[0], skip_special_tokens=True)
     return response
 
-# Create a Gradio interface
+# Create a Gradio interface with a "Generate" button
 iface = gr.Interface(
     fn=generate_response,                     # Function to call for generating response
     inputs=gr.Textbox(label="Input Prompt"),    # Input type (text box for prompt)
     outputs=gr.Textbox(label="Generated Response"),  # Output type (text box for response)
-    live=True                                  # Whether to update output live as user types
+    live=False,                                # Disable live update; only update when button is clicked
+    allow_flagging="never"                     # Prevent flagging (optional, if you don't need it)
 )
 
-# Launch the interface
-iface.launch()
+# Launch the interface with a "Generate" button
+iface.launch(share=True)  # You can set share=True if you want a public link