app.py

import streamlit as st
import torch
from transformers import AutoTokenizer, AutoModelForSequenceClassification
from datasets import load_dataset

# finetuned model
language_model_path = "juliaannjose/finetuned_model"

# load the dataset and
# use the patent number, abstract and claim columns for UI
with st.spinner("Setting up the app..."):
    dataset_dict = load_dataset(
        "HUPD/hupd",
        name="sample",
        data_files="https://huggingface.co/datasets/HUPD/hupd/blob/main/hupd_metadata_2022-02-22.feather",
        icpr_label=None,
        train_filing_start_date="2016-01-01",
        train_filing_end_date="2016-01-21",
        val_filing_start_date="2016-01-22",
        val_filing_end_date="2016-01-31",
    )


# drop down menu with patent numbers
_patent_id = st.selectbox(
    "Select the Patent Number",
    dataset_dict["train"]["patent_number"],
)


# display abstract and claim
@st.cache(persist=True)
def get_abs_claim(_patent_id):
    # get abstract and claim corresponding to this patent id
    _abstract = dataset_dict["train"][["patent_number"] == _patent_id]["abstract"]
    _claim = dataset_dict["train"][["patent_number"] == _patent_id]["claims"]
    return _abstract, _claim


_abstract, _claim = get_abs_claim(_patent_id)
st.write(_abstract)
st.write(_claim)

input_text = _abstract + _claim

# model and tokenizer initialization
tokenizer = AutoTokenizer.from_pretrained(language_model_path)
inputs = tokenizer(
    input_text,
    truncation=True,
    padding=True,
    return_tensors="pt",
)
model = AutoModelForSequenceClassification.from_pretrained(language_model_path)

# get predictions
id2label = {0: "REJECTED", 1: "ACCEPTED"}
# when submit button clicked, run the model and get result
if st.button("Submit"):
    with torch.no_grad():
        logits = model(**inputs).logits

    predicted_class_id = logits.argmax().item()
    pred_label = id2label[predicted_class_id]
    st.write(pred_label)