Update app.py

app.py

@@ -1,26 +1,109 @@
-import numpy as np
+import tensorflow as tf
+import torch
+import torch.nn.functional as F
+from torch.utils.data import Dataset
+#from transformers import BertTokenizer #, BertForSequenceClassification
 import pandas as pd
-from transformers import pipeline
-from transformers import AutoTokenizer, TFAutoModelForSequenceClassification
+import numpy as np
 import streamlit as st
+from sklearn.model_selection import train_test_split
+from transformers import pipeline
+from transformers import AutoTokenizer, AutoModelForSequenceClassification
+from transformers import Trainer, TrainingArguments
+
+#------Test TEXT for ST Forn input
+#I just learned how to suck up to people. You're very good at it, FisherQueen. As for the grammer, it should be  obvious that they're typos, now pick out a mistake here, bitch!
+
+PATH = 'C:/Users/maria/Downloads/bert_base_uncased_fine_tuned_model.pth'
+#saved_model = torch.load(PATH,map_location=torch.device('cpu'))
+
+
+######BERT_MODEL_NAME = 'bert-base-cased'
+BERT_MODEL_NAME = 'distilbert-base-uncased' #'bert-base-uncased'#'bert-base-cased'
+
+#tokenizer = BertTokenizer.from_pretrained(BERT_MODEL_NAME)
+#tokenizer = AutoTokenizer.from_pretrained(BERT_MODEL_NAME)
+###saved_model = torch.load(PATH,map_location=torch.device('cpu'))
+
+LABEL_COLUMNS=["toxic","severe_toxic","obscene","threat","insult","identity_hate"]
+labels = LABEL_COLUMNS
+id2label = {idx:label for idx, label in enumerate(labels)}
+label2id = {label:idx for idx, label in enumerate(labels)}
+
+USER = 'mariasandu/'
+SAVED_MODEL_NAME_ENDING = '-for-toxic-comments-clf'
+
+st.sidebar.header("Choose Model First")
+
+#str = BERT_MODEL_NAME + '-for-toxic-comments-clf'
+model_name = st.sidebar.selectbox("Select Model",
+                                  (
+                                'bert-base-cased' + SAVED_MODEL_NAME_ENDING,
+                                'distilbert-base-uncased' + SAVED_MODEL_NAME_ENDING)
+                                   )
+
+if(model_name == 'bert-base-cased' + SAVED_MODEL_NAME_ENDING):
+      BERT_MODEL_NAME = 'bert-base-cased'
+
+tokenizer = AutoTokenizer.from_pretrained(BERT_MODEL_NAME)
+
+st.sidebar.write('Selected Model:')
+st.sidebar.write(model_name)
+
+saved_model = AutoModelForSequenceClassification.from_pretrained(USER + model_name, #BERT_MODEL_NAME + '-for-toxic-comments-clf',
+                                                                 use_auth_token='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')
+
+def get_text_toxiccom(text):
+  encoding = tokenizer(text, return_tensors="pt")
+  encoding = {k: v.to(saved_model.device) for k,v in encoding.items()}
+
+  outputs = saved_model(**encoding)
+  logits = outputs.logits
+  #print(outputs.logits)
+  #print(logits.shape)
+
+  # apply sigmoid + threshold
+  sigmoid = torch.nn.Sigmoid()
+  probs = sigmoid(logits.squeeze().cpu())
+  pred_prob_list = probs.tolist()
+
+  predictions = np.zeros(probs.shape)
+  predictions[np.where(probs >= 0.5)] = 1
+  # turn predicted id's into actual label names
+  #predicted_labels = [id2label[idx] for idx, label in enumerate(predictions) if label == 1.0]
+  predicted_labels = []
+  for idx,label in enumerate(predictions):
+    if predictions[idx] ==1:
+        predicted_labels.append(labels[idx]) 
+    else:
+        predicted_labels.append('-----')
+   
+  return pred_prob_list,predicted_labels
 
-classifier = pipeline("sentiment-analysis")
 
-model_name ="distilbert-base-uncased-finetuned-sst-2-english"
+st.title('Toxic Comments Application')
+st.write('Welcome to my multi label classification app!')
 
-tokenizer = AutoTokenizer.from_pretrained(model_name)
-model = TFAutoModelForSequenceClassification.from_pretrained(model_name)
+#model_name = st.sidebar.selectbox("Select Model",
+                                  #("distilbert-base-uncased-finetuned-sst-2-english", 
+                                   
+                                   #"finiteautomata/bertweet-base-sentiment-analysis"))
+form = st.form(key='toxic_comments--form')
+user_input = form.text_area('Enter your text')
+submit = form.form_submit_button('Submit')
 
-classifier = pipeline("sentiment-analysis", model=model, tokenizer=tokenizer)
+if submit:
+    text = user_input
+    
+    problst,labellst = get_text_toxiccom(text)
 
+    df = {}
+    df['LABELS'] = LABEL_COLUMNS
+    df['PROBABILITY']= problst
+    df['PREDICTED_LABELS'] = labellst
 
-print(classifier("Thanks a lot for watching the video. Really appreciate it"))
+    outdf = pd.DataFrame.from_dict(df)  #fdict)
+    st.write(outdf)
 
-st.title("Sentiment Analysis of a Tweet")
-st.write("Enter text for sentiment analysis")
-tweet = st.text_input(label="Tweet Text")
 
 
-if(st.button("Analyze")):
-    sentiment = classifier(tweet)
-    st.write(sentiment[0]['label'])