Create inference.py

inference.py

@@ -0,0 +1,31 @@
+import tensorflow as tf
+from tensorflow.keras.models import load_model
+import json
+import keras_nlp
+
+fnet_classifier = load_model("Sentiments classifier.keras")
+
+review_example = input("Input your review: ")
+
+with open("vocab.json", "r") as f:
+    vocab = json.load(f)
+
+seq_max_length = 512
+tokenizer = keras_nlp.tokenizers.WordPieceTokenizer(
+    vocabulary=vocab,
+    lowercase=False,
+    sequence_length=seq_max_length,
+)
+
+def make_prediction(sentence):
+    tokens = tokenizer(review_example)
+    tokens = tf.expand_dims(tokens, 0)
+    prediction = fnet_classifier.predict(tokens, verbose=0)
+
+    if prediction[0][0] > 0.5:
+        result = "The review is POSITIVE"
+    else:
+        result = "The review is NEGATIVE"
+    return result
+result = make_prediction(review_example)
+print(result)