app.py

import gradio as gr

from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline
import torch
from peft import LoraConfig, PeftModel

base_model_name = "microsoft/phi-2"
new_model = "./checkpoint_360"

model = AutoModelForCausalLM.from_pretrained( "microsoft/phi-2", trust_remote_code=True)
model.config.use_cache = False
model.load_adapter(new_model)

tokenizer = AutoTokenizer.from_pretrained(base_model_name, trust_remote_code=True)
tokenizer.pad_token = tokenizer.eos_token
tokenizer.padding_side = "right"

def QLoRA_Chatgpt(prompt):
    print(prompt)
    pipe = pipeline(task="text-generation", model=model, tokenizer=tokenizer, max_length=200)
    result = pipe(f"<s>[INST] {prompt} [/INST]")
    return(result[0]['generated_text'])
    # return "Hello " + name + "!!"

iface = gr.Interface(fn=QLoRA_Chatgpt, inputs=gr.Textbox("how can help you today", label='prompt'), outputs=gr.Textbox(label='Generated-output'))
iface.launch(share=True)