app.py

from transformers import AutoTokenizer, AutoModelForCausalLM

model_name = 'armandnlp/gpt2-TOD_finetuned_SGD'
tokenizer_TOD = AutoTokenizer.from_pretrained(model_name)
model_TOD = AutoModelForCausalLM.from_pretrained(model_name)

def generate_response(message):
    input_ids = tokenizer_TOD(prompt, return_tensors="pt").input_ids
    outputs = model_TOD.generate(input_ids, 
                                 do_sample=False, 
                                 max_length=1024, 
                                 eos_token_id=50262)
    return tokenizer_TOD.batch_decode(outputs)[0]



import gradio as gr

iface = gr.Interface(fn=generate_response, inputs="text", outputs="text")
iface.launch()