Add app.py and weights

app.py

@@ -0,0 +1,92 @@
+import gradio as gr
+import tensorflow as tf
+import re
+from tensorflow import keras
+from tensorflow.keras.preprocessing.sequence import pad_sequences
+from tensorflow.keras.layers import TextVectorization
+import pickle
+import os
+
+os.environ['TF_CPP_MIN_LOG_LEVEL'] = '2'
+
+
+def custom_standardization(input_data):
+    lowercase = tf.strings.lower(input_data)
+    stripped_html = tf.strings.regex_replace(lowercase, "<br />", " ")
+    return tf.strings.regex_replace(
+        stripped_html, "[%s]" % re.escape("!#$%&'()*+,-./:;<=>?@\^_`{|}~"), ""
+    )
+
+
+count_vect = pickle.load(open('countvect.pkl', 'rb'))
+tokenizer = pickle.load(open('tokenizer.pkl', 'rb'))
+
+from_disk = pickle.load(open('tv_layer.pkl', 'rb'))
+text_vectorization = TextVectorization.from_config(from_disk['config'])
+text_vectorization.set_weights(from_disk['weights'])
+
+lr_model = pickle.load(open('logistic_model.pkl', 'rb'))
+lstm_model = keras.models.load_model('lstm_model.h5')
+bert_classifier_model = keras.models.load_model('bert_classifier.h5')
+
+
+def get_bert_end_to_end(model):
+    inputs_string = keras.Input(shape=(1,), dtype="string")
+    indices = text_vectorization(inputs_string)
+    outputs = model(indices)
+    end_to_end_model = keras.Model(inputs_string, outputs, name="end_to_end_model")
+    optimizer = keras.optimizers.Adam(learning_rate=0.001)
+    end_to_end_model.compile(
+        optimizer=optimizer, loss="binary_crossentropy", metrics=["accuracy"]
+    )
+    return end_to_end_model
+
+
+bert_end_model = get_bert_end_to_end(bert_classifier_model)
+
+
+def get_lr_results(text):
+    sample_vec = count_vect.transform([text])
+    return lr_model.predict(sample_vec)[0]
+
+
+def get_lstm_results(text):
+    tokenized_text = tokenizer.texts_to_sequences([text])
+    padded_tokens = pad_sequences(tokenized_text, maxlen=200)
+    return lstm_model.predict(padded_tokens)[0][0]
+
+
+def get_bert_results(text):
+    return bert_end_model.predict([text])[0][0]
+
+
+def decide(text):
+    lr_result = get_lr_results(text)
+    lstm_result = get_lstm_results(text)
+    bert_result = get_bert_results(text)
+    results = [
+        lr_result.round(2),
+        lstm_result.round(2),
+        bert_result.round(2)]
+    if lstm_result >= 0.6:
+        return "Positive review (LR: {}, LSTM: {}, BERT: {}".format(*results)
+    elif lstm_result <= 0.4:
+        return "Negative review (LR: {}, LSTM: {}, BERT: {}".format(*results)
+    else:
+        return "Neutral review (LR: {}, LSTM: {}, BERT: {}".format(*results)
+
+
+example_sentence_1 = "I hate this toaster, they made no effort in making it. So cheap, it almost immediately broke!"
+example_sentence_2 = "Great toaster! We love the way it toasted my bread so quickly. Very high quality components too."
+examples = [[example_sentence_1], [example_sentence_2]]
+
+description = "Write out a product review to know the underlying sentiment."
+
+gr.Interface(decide,
+             inputs=gr.inputs.Textbox(lines=1, placeholder=None, default="", label=None),
+             outputs='text',
+             examples=examples,
+             title="Sentiment analysis of product reviews",
+             theme="grass", description=description,
+             allow_flagging="auto",
+             flagging_dir='flagging records').launch(enable_queue=True, inline=False, share=True)

bert_classifier.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:40773586c34fe1c3197640db0267716bedb5f76e1ffebb6b7232806741452178
+size 16501864

countvect.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:117ac083bd1587e1cba48feb9669bb7e5e0871846a497c414e51e9610de8d946
+size 14439392

logistic_model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:93bf191cb6320e2081ee5cf4ee695a497b4165f398e3ef877401d5787f55576d
+size 7016090

lstm_model.h5

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:ee86ce5eac1a091e8997adf34c2b056769ae482f99cffa82c4274e5f4179b193
+size 11066736

tokenizer.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4de6458d9a8021512a065ed2d64d182289fcb1333aa17e34f8c12a5c5f7cb222
+size 49477377

tv_layer.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:fe6f5c5513bb7515866c9bf81c4a51438e9da90963414f7eb56f68a701e50ab4
+size 298869