app.py

import gradio as gr

import model_utils
import inference

tokenizer = model_utils.load_tokenizers()
tokenizer.pad_token = tokenizer.eos_token

model = model_utils.load_model(quantization_config=None)
adapter = "model/checkpoint-700/"
model.load_adapter(adapter)

def generated_text(prompt):
    result = inference.predict(prompt=prompt, model=model, tokenizer=tokenizer, max_length=200)
    return result

iface = gr.Interface(
    fn = generated_text,
    inputs = gr.Textbox(),
    outputs = gr.Textbox(),
    theme="dark",
    title="Phi-2 Finetuned Model. For more visit: https://github.com/bala1802/Phi2"
)

iface.launch(share=True)