Update app.py

app.py

@@ -1,133 +1,23 @@
-# -*- coding: utf-8 -*-
-"""First_Text_Classification.ipynb
-
-Automatically generated by Colaboratory.
-
-Original file is located at
-    https://colab.research.google.com/drive/1sdLss09e3OxYVoeK3oBA6qrUSj_iOxp-
-
-<h3 align = "center">Importing Libraries</h3>
-"""
-
-import numpy as np
-import pandas as pd
-
-"""<h3 align = "center">Importing Dataset</h3>"""
-
-data = pd.read_csv("spam.csv", encoding = "ISO-8859-1")
-
-"""<h3 align = "center">Preliminary Data Checks</h3>"""
-
-data.head()
-
-data.isnull().sum()
-
-data.shape
-
-data['v1'].value_counts()
-
-data.info()
-
-"""<h3 align = "center">Putting the Length of Characters of each row in a column.</h3>"""
-
-data["Unnamed: 2"] = data["v2"].str.len()
-
-"""<h3 align = "center">Visualising Length of Characters for each category!</h3>"""
-
-
-"""<h5>It is evident from the above plot that spam texts are usually longer in length!</h5>
-
-<h3 align = "center">Defining Variables</h3>
-"""
-
-X = data["v2"]
-y = data["v1"]
-
-"""<h3 align = "center">Train Test Split</h3>"""
-
-from sklearn.model_selection import train_test_split
-X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.33, random_state=42)
-
-"""<h3 align = "center">Vecrorizing Words into Matrix</h3>"""
-
-from sklearn.feature_extraction.text import CountVectorizer
-count_vect = CountVectorizer()
-
-X_train_counts = count_vect.fit_transform(X_train)
-
-X_train_counts
-
-X_train.shape
-
-X_train_counts.shape
-
-from sklearn.feature_extraction.text import TfidfTransformer
-tfidf_transformer = TfidfTransformer()
-
-X_train_tfidf = tfidf_transformer.fit_transform(X_train_counts)
-
-X_train_tfidf.shape
-
-"""<h3 align = "center">Using TDIF Vectorizer for optimum vectorization!</h3>"""
-
-from sklearn.feature_extraction.text import TfidfVectorizer
-vectorizer = TfidfVectorizer()
-
-X_train_tfidf = vectorizer.fit_transform(X_train)
-
-X_train_tfidf.shape
-
-"""<h3 align = "center">Creating Model</h3>"""
-
-from sklearn.svm import LinearSVC
-clf = LinearSVC()
-
-clf.fit(X_train_tfidf,y_train)
-
-"""<h3 align = "center">Creating Pipeline</h3>"""
-
-from sklearn.pipeline import Pipeline
-
-text_clf = Pipeline([("tfidf",TfidfVectorizer()),("clf",LinearSVC())])
-
-text_clf.fit(X_train,y_train)
-
-predictions = text_clf.predict(X_test)
-
-X_test
-
-from sklearn.metrics import confusion_matrix,classification_report,accuracy_score
-
-print(confusion_matrix(y_test,predictions))
-
-print(classification_report(y_test,predictions))
-
-"""<h3 align = "center">Accuracy Score</h3>"""
-
-print(accuracy_score(y_test,predictions))
-
-"""<h3 align = "center">Predictions </h3>"""
-
-text_clf.predict(["Hi how are you doing today?"])
-
-text_clf.predict(["Congratulations! You are selected for a free vouchar worth $500"])
-
-"""<h3 align = "center">Creating User Interface!</h3>"""
-
+import os
 import gradio as gr
+from transformers import AutoTokenizer, TFAutoModelForSequenceClassification
+
+HF_TOKEN = os.environ.get('HF_TOKEN')
 
-def first_nlp_spam_detector(text):
-  list = []
-  list.append(text)
-  arr =  text_clf.predict(list)
-  if arr[0] == 'ham':
-    return "Your Text is a Legitimate One!"
-  else:
-    return "Beware of such text messages, It\'s a Spam! "
+model_checkpoint = "besijar/dspa_review_classification"
+tokeniser = AutoTokenizer.from_pretrained(model_checkpoint, use_auth_token=HF_TOKEN)
+model = TFAutoModelForSequenceClassification.from_pretrained(model_checkpoint, use_auth_token=HF_TOKEN)
 
-interface = gr.Interface(first_nlp_spam_detector,inputs = gr.Textbox(lines=2, placeholder="Enter your Text Here.....!", show_label = False),
-                         outputs = gr.Label(value = "Predicting the Text Classification..!"),description = "Predicting Text Legitimacy!")
+example_review = "Tully's House Blend is the perfect K-Cup for me.  Sure, I occasionally enjoy the special flavors.....Mocha, Italian roast, French vanilla, but my favorite 'go-to'coffee is House Blend. Wakes me up in the morning with it's coffee house full hearty taste."
 
-first_nlp_spam_detector("Congratulations! You are selected for a free vouchar worth $500")
+def review_classify(review):
+    review = tokeniser.encode(review)
+    review = model.predict([review])
+    return int(review.logits.argmax())
 
-interface.launch()
+iface = gr.Interface(review_classify,
+                        title="Review Classification using DistilRoBERTa",
+                        inputs=[gr.Text(label="Review")],
+                        outputs=[gr.Number(label="Rating", precision=0)],
+                        examples=[example_review])
+iface.launch()