|
import streamlit as st |
|
import pandas as pd |
|
import numpy as np |
|
from transformers import pipeline, AutoModelWithLMHead, AutoTokenizer |
|
import torch |
|
import boto3 |
|
|
|
s3 = boto3.resource('s3') |
|
s3_object = s3.Bucket('nlp-gpt-models').Object('mod_v1.pth').get() |
|
|
|
|
|
|
|
st.title('Patent Context Generation Tool-Development Stage..') |
|
model_path = s3_object |
|
|
|
|
|
propmt_title = st.text_input('Enter Your Title....', 'Biology...') |
|
|
|
f1 = st.button('Generate') |
|
if f1: |
|
try: |
|
saved_model = torch.load(model_path) |
|
tokenizer = AutoTokenizer.from_pretrained("gpt2") |
|
generator = pipeline('text-generation', model = saved_model , tokenizer = tokenizer) |
|
def paraphrase(propmt_title): |
|
p = generator('<s>' + propmt_title + '</s>>>>><p>') |
|
return p[0]['generated_text'].split('</s>>>>>><p>')[0].split('</p>')[0].split('<p>')[1] |
|
output= paraphrase(propmt_title) |
|
st.text_area('paraphrased_titless', output ,False) |
|
except Exception as e: |
|
st.exception("Exception: %s\n" % e) |
|
st.text_area('paraphrased_titless', st.exception("Exception: %s\n" % e) ,False) |
|
|
|
propmt_title = st.text_input('Enter Your Paraphrased Title....', 'title context...') |
|
f2 = st.form("my_form2") |
|
f2.form_submit_button("Submit") |
|
|
|
st.text_area('Generated Fields', '',False) |
|
|
|
|
|
propmt_title = st.text_input('Enter Your Generated Field ....', 'Field context...') |
|
f3 = st.form("my_form3") |
|
f3.form_submit_button("Submit") |
|
|
|
st.text_area('Generated Abstract', '',False) |
|
|
|
|
|
propmt_title = st.text_input('Enter Your Generated Abstract ....', 'Abstract context...') |
|
f4 = st.form("my_form4") |
|
f4.form_submit_button("Submit") |
|
|
|
st.text_area('Generated Background', '',False) |
|
|
|
|
|
|
|
st.download_button(label="Download Full Patent (PDF file)", |
|
data=propmt_title, |
|
file_name="test.pdf", |
|
mime='application/octet-stream') |
