app.py

import torch
from peft import PeftModel, PeftConfig
from transformers import AutoModelForCausalLM, AutoTokenizer

peft_model_id = f"TCFSBN/egrantha-keyword-inference"
config = PeftConfig.from_pretrained(peft_model_id)
model = AutoModelForCausalLM.from_pretrained(
    config.base_model_name_or_path,
    return_dict=True,
    load_in_8bit=True,
    device_map="auto",
)
tokenizer = AutoTokenizer.from_pretrained(config.base_model_name_or_path)

# Load the Lora model
model = PeftModel.from_pretrained(model, peft_model_id)


def make_inference(abstract):
    input_str = (f"Below is a technical abstract or needs statement, please list keywords relevant to the technical abstract " +
                 f"or needs statement.\n\n### Technical abstract/needs statement:\n{abstract}\n### Keywords:\n")
    batch = tokenizer(input_str, return_tensors="pt")

    with torch.cuda.amp.autocast():
        output_tokens = model.generate(**batch, max_new_tokens=50)

    return tokenizer.decode(output_tokens[0], skip_special_tokens=True)


if __name__ == "__main__":
    # make a gradio interface
    import gradio as gr

    gr.Interface(
        make_inference,
        [
            gr.inputs.Textbox(lines=20, label="Abstract"),

        ],
        gr.outputs.Textbox(label="Keywords"),
        title="arxiv-inference",
        description="Try to generate a set of keywords from an abstract.",
    ).launch()