| from transformers import AutoTokenizer | |
| import streamlit as st | |
| from transformers import pipeline | |
| from transformers import ( | |
| TFAutoModelForSequenceClassification as AutoModelForSequenceClassification, | |
| ) | |
| st.title("Detecting Toxic Tweets") | |
| demo = """I'm so proud of myself for accomplishing my goals today. #motivation #success""" | |
| text = st.text_area("Input text", demo, height=250) | |
| model_name = "distilbert-base-uncased-finetuned-sst-2-english" | |
| tokenizer = AutoTokenizer.from_pretrained(model_name) | |
| model = AutoModelForSequenceClassification.from_pretrained(model_name) | |
| clf = pipeline( | |
| "toxicity-analysis", model=model, tokenizer=tokenizer, return_all_scores=True | |
| ) | |
| input = tokenizer(text, return_tensors="tf") | |
| if st.button("Submit", type="primary"): | |
| results = clf(text)[0] | |
| classes = dict(d.values() for d in results) | |
| st.bar_chart(classes) | |