quick_ref.txt

https://huggingface.co./spaces/juliaannjose/hupd_patent_classifier/


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


# finetuned model
language_model_path = "juliaannjose/finetuned_model"

# load the dataset to
# use the patent number, abstract and claim columns for UI
with st.spinner("Loading..."):
    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",
    )
    df_train = pd.DataFrame(dataset_dict["train"])
    df_val = pd.DataFrame(dataset_dict["validation"])
    df = pd.concat([df_train, df_val], ignore_index=True)


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


# display abstract and claim
def get_abs_claim(_pid):
    # get abstract and claim corresponding to this patent id
    _abs = df.loc[df["patent_number"] == _pid]["abstract"]
    _cl = df.loc[df["patent_number"] == _pid]["claims"]
    return _abs.values[0], _cl.values[0]


_abstract, _claim = get_abs_claim(_patent_id)
st.title("Abstract:")  # display abstract
st.write(_abstract)
st.title("Claim:")  # display claims
st.write(_claim)


# model and tokenizer initialization
@st.cache_resource
def load_model(language_model_path):
    tokenizer = AutoTokenizer.from_pretrained(language_model_path)
    model = AutoModelForSequenceClassification.from_pretrained(language_model_path)
    return tokenizer, model


tokenizer, model = load_model(language_model_path)
# input to our model
input_text = _abstract + _claim
# get tokens
inputs = tokenizer(
    input_text,
    truncation=True,
    padding=True,
    return_tensors="pt",
)

# 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():
        outputs = model(**inputs)
        probability = torch.nn.functional.softmax(outputs.logits, dim=1)

    predicted_class_id = probability.argmax().item()
    pred_label = id2label[predicted_class_id]
    st.title("Predicted Patentability")
    if probability[0][0] > probability[0][1]:
        st.write("Rejection Score:")
        st.write(probability[0][0].item())
    else:
        st.write("Acceptance Score:")
        st.write(probability[0][1].item())
    st.write("Result:", pred_label)





-----------------------------------------------------------