Spaces:
Runtime error
Runtime error
| import numpy as np | |
| import pandas as pd | |
| import streamlit as st | |
| from streamlit_text_rating.st_text_rater import st_text_rater | |
| from transformers import AutoTokenizer,AutoModelForSequenceClassification | |
| import onnxruntime as ort | |
| import os | |
| import time | |
| import plotly.express as px | |
| import plotly.graph_objects as go | |
| global _plotly_config | |
| _plotly_config={'displayModeBar': False} | |
| from sentiment_clf_helper import classify_sentiment,create_onnx_model_sentiment,classify_sentiment_onnx | |
| from zeroshot_clf_helper import zero_shot_classification,create_onnx_model_zs,zero_shot_classification_onnx | |
| import yaml | |
| def read_yaml(file_path): | |
| with open(file_path, "r") as f: | |
| return yaml.safe_load(f) | |
| config = read_yaml('config.yaml') | |
| sent_chkpt=config['SENTIMENT_CLF']['sent_chkpt'] | |
| sent_mdl_dir=config['SENTIMENT_CLF']['sent_mdl_dir'] | |
| sent_onnx_mdl_dir=config['SENTIMENT_CLF']['sent_onnx_mdl_dir'] | |
| sent_onnx_mdl_name=config['SENTIMENT_CLF']['sent_onnx_mdl_name'] | |
| sent_onnx_quant_mdl_name=config['SENTIMENT_CLF']['sent_onnx_quant_mdl_name'] | |
| zs_chkpt=config['ZEROSHOT_CLF']['zs_chkpt'] | |
| zs_mdl_dir=config['ZEROSHOT_CLF']['zs_mdl_dir'] | |
| zs_onnx_mdl_dir=config['ZEROSHOT_CLF']['zs_onnx_mdl_dir'] | |
| zs_onnx_mdl_name=config['ZEROSHOT_CLF']['zs_onnx_mdl_name'] | |
| zs_onnx_quant_mdl_name=config['ZEROSHOT_CLF']['zs_onnx_quant_mdl_name'] | |
| st.set_page_config( # Alternate names: setup_page, page, layout | |
| layout="wide", # Can be "centered" or "wide". In the future also "dashboard", etc. | |
| initial_sidebar_state="auto", # Can be "auto", "expanded", "collapsed" | |
| page_title='None', # String or None. Strings get appended with "• Streamlit". | |
| ) | |
| padding_top = 0 | |
| st.markdown(f""" | |
| <style> | |
| .reportview-container .main .block-container{{ | |
| padding-top: {padding_top}rem; | |
| }} | |
| </style>""", | |
| unsafe_allow_html=True, | |
| ) | |
| def set_page_title(title): | |
| st.sidebar.markdown(unsafe_allow_html=True, body=f""" | |
| <iframe height=0 srcdoc="<script> | |
| const title = window.parent.document.querySelector('title') \ | |
| const oldObserver = window.parent.titleObserver | |
| if (oldObserver) {{ | |
| oldObserver.disconnect() | |
| }} \ | |
| const newObserver = new MutationObserver(function(mutations) {{ | |
| const target = mutations[0].target | |
| if (target.text !== '{title}') {{ | |
| target.text = '{title}' | |
| }} | |
| }}) \ | |
| newObserver.observe(title, {{ childList: true }}) | |
| window.parent.titleObserver = newObserver \ | |
| title.text = '{title}' | |
| </script>" /> | |
| """) | |
| set_page_title('NLP use cases') | |
| # Hide Menu Option | |
| hide_streamlit_style = """ | |
| <style> | |
| #MainMenu {visibility: hidden;} | |
| footer {visibility: hidden;} | |
| </style> | |
| """ | |
| st.markdown(hide_streamlit_style, unsafe_allow_html=True) | |
| def create_model_dir(chkpt, model_dir): | |
| if not os.path.exists(model_dir): | |
| try: | |
| os.mkdir(path=model_dir) | |
| except: | |
| pass | |
| _model = AutoModelForSequenceClassification.from_pretrained(chkpt) | |
| _tokenizer = AutoTokenizer.from_pretrained(chkpt) | |
| _model.save_pretrained(model_dir) | |
| _tokenizer.save_pretrained(model_dir) | |
| else: | |
| pass | |
| #title using markdown | |
| st.markdown("<h1 style='text-align: center; color: #3366ff;'>NLP Basic Use Cases</h1>", unsafe_allow_html=True) | |
| st.markdown("---") | |
| with st.sidebar: | |
| # title using markdown | |
| st.markdown("<h1 style='text-align: left; color: ;'>NLP Tasks</h1>", unsafe_allow_html=True) | |
| select_task=st.selectbox(label="Select task from drop down menu", | |
| options=['README', | |
| 'Detect Sentiment','Zero Shot Classification']) | |
| ############### Pre-Download & instantiate objects for sentiment analysis *********************** START ********************** | |
| # #create model/token dir for sentiment classification for faster inference | |
| create_model_dir(chkpt=sent_chkpt, model_dir=sent_mdl_dir) | |
| def sentiment_task_selected(task, | |
| sent_chkpt=sent_chkpt, | |
| sent_mdl_dir=sent_mdl_dir, | |
| sent_onnx_mdl_dir=sent_onnx_mdl_dir, | |
| sent_onnx_mdl_name=sent_onnx_mdl_name, | |
| sent_onnx_quant_mdl_name=sent_onnx_quant_mdl_name): | |
| #model & tokenizer initialization for normal sentiment classification | |
| # model_sentiment=AutoModelForSequenceClassification.from_pretrained(sent_chkpt) | |
| # tokenizer_sentiment=AutoTokenizer.from_pretrained(sent_chkpt) | |
| tokenizer_sentiment = AutoTokenizer.from_pretrained(sent_mdl_dir) | |
| # # create onnx model for sentiment classification but once created in your local app comment this out | |
| # create_onnx_model_sentiment(_model=model_sentiment, _tokenizer=tokenizer_sentiment) | |
| #create inference session | |
| sentiment_session = ort.InferenceSession(f"{sent_onnx_mdl_dir}/{sent_onnx_mdl_name}") | |
| # sentiment_session_quant = ort.InferenceSession(f"{sent_onnx_mdl_dir}/{sent_onnx_quant_mdl_name}") | |
| return tokenizer_sentiment,sentiment_session | |
| ############## Pre-Download & instantiate objects for sentiment analysis ********************* END ********************************** | |
| ############### Pre-Download & instantiate objects for Zero shot clf *********************** START ********************** | |
| # create model/token dir for zeroshot clf -- already created so not required | |
| create_model_dir(chkpt=zs_chkpt, model_dir=zs_mdl_dir) | |
| def zs_task_selected(task, | |
| zs_chkpt=zs_chkpt , | |
| zs_mdl_dir=zs_mdl_dir, | |
| zs_onnx_mdl_dir=zs_onnx_mdl_dir, | |
| zs_onnx_mdl_name=zs_onnx_mdl_name, | |
| zs_onnx_quant_mdl_name=zs_onnx_quant_mdl_name): | |
| ##model & tokenizer initialization for normal ZS classification | |
| # model_zs=AutoModelForSequenceClassification.from_pretrained(zs_chkpt) | |
| # we just need tokenizer for inference and not model since onnx model is already saved | |
| # tokenizer_zs=AutoTokenizer.from_pretrained(zs_chkpt) | |
| tokenizer_zs = AutoTokenizer.from_pretrained(zs_mdl_dir) | |
| # # create onnx model for zeroshot but once created locally comment it out. | |
| # create_onnx_model_zs() | |
| #create inference session from onnx model | |
| zs_session = ort.InferenceSession(f"{zs_onnx_mdl_dir}/{zs_onnx_mdl_name}") | |
| # zs_session_quant = ort.InferenceSession(f"{zs_onnx_mdl_dir}/{zs_onnx_quant_mdl_name}") | |
| return tokenizer_zs,zs_session | |
| ############## Pre-Download & instantiate objects for Zero shot analysis ********************* END ********************************** | |
| if select_task=='README': | |
| st.header("NLP Summary") | |
| # st.write() | |
| if select_task == 'Detect Sentiment': | |
| t1=time.time() | |
| tokenizer_sentiment,sentiment_session = sentiment_task_selected(task=select_task) | |
| t2 = time.time() | |
| st.write(f"Total time to load Model is {(t2-t1)*1000:.1f} ms") | |
| st.header("You are now performing Sentiment Analysis") | |
| input_texts = st.text_input(label="Input texts separated by comma") | |
| c1,c2,_,_=st.columns(4) | |
| with c1: | |
| response1=st.button("Compute (ONNX runtime)") | |
| if response1: | |
| start = time.time() | |
| sentiments=classify_sentiment_onnx(input_texts, | |
| _session=sentiment_session, | |
| _tokenizer=tokenizer_sentiment) | |
| end = time.time() | |
| st.write(f"Time taken for computation {(end - start) * 1000:.1f} ms") | |
| for i,t in enumerate(input_texts.split(',')): | |
| if sentiments[i]=='Positive': | |
| response=st_text_rater(t + f"--> This statement is {sentiments[i]}", | |
| color_background='rgb(154,205,50)',key=t) | |
| else: | |
| response = st_text_rater(t + f"--> This statement is {sentiments[i]}", | |
| color_background='rgb(233, 116, 81)',key=t) | |
| else: | |
| pass | |
| if select_task=='Zero Shot Classification': | |
| t1=time.time() | |
| tokenizer_zs,zs_session = zs_task_selected(task=select_task) | |
| t2 = time.time() | |
| st.write(f"Total time to load Model is {(t2-t1)*1000:.1f} ms") | |
| st.header("You are now performing Zero Shot Classification") | |
| input_texts = st.text_input(label="Input text to classify into topics") | |
| input_lables = st.text_input(label="Enter labels separated by commas") | |
| c1,_,_,_=st.columns(4) | |
| with c1: | |
| response1=st.button("Compute (ONNX runtime)") | |
| if response1: | |
| start = time.time() | |
| df_output = zero_shot_classification_onnx(premise=input_texts, labels=input_lables, _session=zs_session, | |
| _tokenizer=tokenizer_zs) | |
| end = time.time() | |
| st.write("") | |
| st.write(f"Time taken for computation {(end-start)*1000:.1f} ms") | |
| fig = px.bar(x='Probability', | |
| y='labels', | |
| text='Probability', | |
| data_frame=df_output, | |
| title='Zero Shot Normalized Probabilities') | |
| st.plotly_chart(fig, config=_plotly_config) | |
| else: | |
| pass | |