Transformer-Encoder-with-Char

1. Transformer Encoder with Char information for text classification
2. This code was created by referring to the code in carpedm20 and DongjunLee

1. Model structure

1. Input words are represented with Char-CNN, Word2vec concatenated together(64 dimensions each)

2. Normal Transformer Encoder from (Attention is all you need) is used

3. Model is composed of 7 Transformer Encoder layers with 4 attention heads

4. Global Average Pooling layer with softmax is used at the end, for predicting class

2. Char CNN

1. Char CNN implemented by Yoon Kim

3. Prerequisite

• Tensorflow 1.8.0
• Python 3.6

4. Training

1. Clone git

$ git clone https://github.com/MSWon/Transformer-Encoder-with-Char.git

2. Unzip data.zip and embedding.zip

$ unzip data.zip
$ unzip embedding.zip

3. Training with user settings (char_mode : (char_cnn, char_lstm, no_char))

$ python train.py --batch_size 128 --training_epochs 12 --char_mode char_cnn

5. Experiments

5-1. Datasets

1. The AG’s news topic classification dataset is constructed by choosing 4 largest classes from the original news corpus
2. 4 classes are ‘world’, ‘sports’, ‘business’ and ‘science/technology’
3. Each class contains 30,000 training samples and 1,900 testing samples
4. The total number of training samples is 120,000 and 7,600 for test

5-2. Test loss graph

5-3. Performance table