train.py

# libraries
import random
from tensorflow.keras.optimizers import SGD
from keras.layers import Dense, Dropout
from keras.models import load_model
from keras.models import Sequential
import numpy as np
import pickle
import json
import nltk
from nltk.stem import WordNetLemmatizer
lemmatizer = WordNetLemmatizer()
nltk.download('omw-1.4')
nltk.download("punkt")
nltk.download("wordnet")


# init file
words = []
classes = []
documents = []
ignore_words = ["?", "!"]
data_file = open("/home/quangjimmy/chatbotmini/intents.json").read()
intents = json.loads(data_file)

# words
for intent in intents["intents"]:
    for pattern in intent["patterns"]:

        # take each word and tokenize it
        w = nltk.word_tokenize(pattern)
        words.extend(w)
        # adding documents
        documents.append((w, intent["tag"]))

        # adding classes to our class list
        if intent["tag"] not in classes:
            classes.append(intent["tag"])

# lemmatizer
words = [lemmatizer.lemmatize(w.lower()) for w in words if w not in ignore_words]
words = sorted(list(set(words)))

classes = sorted(list(set(classes)))

print(len(documents), "documents")

print(len(classes), "classes", classes)

print(len(words), "unique lemmatized words", words)


pickle.dump(words, open("words.pkl", "wb"))
pickle.dump(classes, open("classes.pkl", "wb"))

# training initializer
# initializing training data
training = []
output_empty = [0] * len(classes)
for doc in documents:
    # initializing bag of words
    bag = []
    # list of tokenized words for the pattern
    pattern_words = doc[0]
    # lemmatize each word - create base word, in attempt to represent related words
    pattern_words = [lemmatizer.lemmatize(word.lower()) for word in pattern_words]
    # create our bag of words array with 1, if word match found in current pattern
    for w in words:
        bag.append(1) if w in pattern_words else bag.append(0)

    # output is a '0' for each tag and '1' for current tag (for each pattern)
    output_row = list(output_empty)
    output_row[classes.index(doc[1])] = 1

    training.append([bag, output_row])

# shuffle our features and turn into np.array
random.shuffle(training)

# training = np.array(training)
# # create train and test lists. X - patterns, Y - intents
# train_x = list(training[:, 0])
# train_y = list(training[:, 1])

#updated

# Separate bag-of-words representations and output labels
train_x = [item[0] for item in training]
train_y = [item[1] for item in training]

# Convert to NumPy arrays
train_x = np.array(train_x)
train_y = np.array(train_y)
print("Training data created")

# actual training
# Create model - 3 layers. First layer 128 neurons, second layer 64 neurons and 3rd output layer contains number of neurons
# equal to number of intents to predict output intent with softmax
model = Sequential()
model.add(Dense(128, input_shape=(len(train_x[0]),), activation="relu"))
model.add(Dropout(0.5))
model.add(Dense(64, activation="relu"))
model.add(Dropout(0.5))
model.add(Dense(len(train_y[0]), activation="softmax"))
model.summary()

# Compile model. Stochastic gradient descent with Nesterov accelerated gradient gives good results for this model

# sgd = SGD(lr=0.01, decay=1e-6, momentum=0.9, nesterov=True)
# model.compile(loss="categorical_crossentropy", optimizer=sgd, metrics=["accuracy"])

#Updated (Removed decayIt seems like you're using a deprecated argument, decay, in the instantiation of the SGD optimizer from Keras. The decay argument has been deprecated in newer versions of Keras. To address this issue, 
# you can switch to using the newer format for specifying learning rate schedules in the optimizer.)

sgd = SGD(learning_rate=0.01, momentum=0.9, nesterov=True)
model.compile(loss="categorical_crossentropy", optimizer=sgd, metrics=["accuracy"])


# for choosing an optimal number of training epochs to avoid underfitting or overfitting use an early stopping callback to keras
# based on either accuracy or loos monitoring. If the loss is being monitored, training comes to halt when there is an 
# increment observed in loss values. Or, If accuracy is being monitored, training comes to halt when there is decrement observed in accuracy values.

# from keras import callbacks 
# earlystopping = callbacks.EarlyStopping(monitor ="loss", mode ="min", patience = 5, restore_best_weights = True)
# callbacks =[earlystopping]

# fitting and saving the model
hist = model.fit(np.array(train_x), np.array(train_y), epochs=200, batch_size=5, verbose=1)
model.save("chatbot_model.h5", hist)
print("model created")