app.py

import gradio as gr
import transformers

# Load a pre-trained model.
model = transformers.AutoModelForSeq2SeqLM.from_pretrained("facebook/bart-large")

# Define a function to generate text.
def generate_text(text):
  """Generates text based on a given prompt."""

  # Tokenize the input text.
  input_ids = model.tokenizer.encode(text, return_tensors="pt")

  # Generate text.
  output_ids = model.generate(input_ids=input_ids, max_length=100, num_beams=5)

  # Decode the output text.
  output_text = model.tokenizer.decode(output_ids[0])

  return output_text

# Define the Gradio interface.
chat_box = gr.inputs.Textbox(label="Chat Box")
chat_button = gr.Button("Send")
chat_response = gr.outputs.Textbox(label="Chat Response")

# Connect the inputs and outputs to the generate_text function.
chat_button.click(generate_text, chat_box, chat_response)

# Launch the Gradio interface.
interface = gr.Interface([chat_box, chat_button], [chat_response])
interface.launch()