Skip to content

Latest commit

 

History

History
111 lines (78 loc) · 3.76 KB

Tutorial_2_Classification.md

File metadata and controls

111 lines (78 loc) · 3.76 KB
Raw

Tutorial 2: Text Classification

Kashgari provides CNN_Model, CNN_LSTM_Model and BLSTM_Model for text classification, All classification models inherit from the ClassificationModel. You could easily switch from one model to another just by changing one line of code.

Also, it is very easy to customize your own model by using the ClassificationModel class. And all the models could build with Word2vec embedding, BERT embedding or just an empty embedding layer.

Here is a real-life example of how to train and save on your own corpus.

from kashgari.utils.logger import init_logger
init_logger()

from kashgari.tasks.classification import CNNLSTMModel

# prepare corpus
train_x = [
            list('语言学（英语：linguistics）是一门关于人类语言的科学研究'),
            list('语言学（英语：linguistics）是一门关于人类语言的科学研究'),
            list('语言学（英语：linguistics）是一门关于人类语言的科学研究'),
            list('语言学包含了几种分支领域。'),
            list('在语言结构（语法）研究与意义（语义与语用）研究之间存在一个重要的主题划分'),
        ]
train_y = ['a', 'a', 'a', 'b', 'c']

# train model
model = CNNLSTMModel()
model.fit(train_x, train_y)

# save model
model.save('./some-where-some-where/')

Load trained model

from kashgari.tasks.classification import ClassificationModel

model = ClassificationModel.load_model('./some-where-some-where/')
model.predict(list('加载完成咯'))

Adjust models hyper params

from kashgari.tasks.classification import CNNLSTMModel
hyper_params = CNNLSTMModel.base_hyper_parameters.copy()
print(hyper_params)
# {'conv_layer': {'filters': 32, 'kernel_size': 3, 'padding': 'same', 'activation': 'relu'}, 'max_pool_layer': {'pool_size': 2}, 'lstm_layer': {'units': 100}}

hyper_params['lstm_layer']['units'] = 300
model = CNNLSTMModel(hyper_parameters=hyper_params)

Customize your own model

It is very easy and straightforward to build your own customized model, just inherit the ClassificationModel and implement the build_model() function.

from keras.layers import LSTM, Dense, Dropout
from keras.models import Model
from kashgari.tasks.classification import ClassificationModel


class MyOwnModel(ClassificationModel):
    __architect_name__ = 'MyOwnModel'

    def build_model(self):
        base_model = self.embedding.model
        lstm_layer = LSTM(100, return_sequences=True)(base_model.output)
        drop_out_layer = Dropout(0.3)(lstm_layer)
        dense_layer = Dense(len(self.label2idx), activation='sigmoid')(drop_out_layer)
        output_layers = [dense_layer]

        model = Model(base_model.inputs, output_layers)
        model.compile(loss='categorical_crossentropy',
                      optimizer='adam',
                      metrics=['accuracy'])
        self.model = model
        self.model.summary()

model = MyOwnModel()
model.fit(train_x, train_y)

model.save('./my_own_model')

loaded_model = MyOwnModel.load_model('./my_own_model')

Use pre-embedding embedding layer

see Tutorial 1: Word Embeddings for detail. All models, include the customized one support using word2vec or BERT embedding as the embedding layer.

Use callbacks

Kashgari is based on keras so that you could use all of the keras callbacks directly with Kashgari model. For example, here is how to visualize training with tensorboard.

import keras

tf_board_callback = keras.callbacks.TensorBoard(log_dir='./logs', update_freq=1000)


model = CNNModel()
model.fit(train_x,
          train_y,
          val_x,
          val_y,
          batch_size=100,
          fit_kwargs={'callbacks': [tf_board_callback]})