app.py

import torch
from transformers import AutoModelForCausalLM, AutoTokenizer, pipeline, logging
import gradio as gr


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


tokenizer = AutoTokenizer.from_pretrained(model_name, trust_remote_code=True)
tokenizer.pad_token = tokenizer.eos_token

# Loading adapter (trained LORA weights)
ckpt = '/content/drive/MyDrive/S27/results/checkpoint-500'
model.load_adapter(ckpt)
# adapter_path = 'checkpoint-500'
# model.load_adapter(adapter_path)

def inference(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']
    
with gr.Blocks() as demo:
    prompt = gr.Textbox(label="Prompt")
    output = gr.Textbox(label="Output Box")
    greet_btn = gr.Button("Generate")
    greet_btn.click(fn=inference, inputs=prompt, outputs=output, api_name="inference")

demo.launch()