An implementation of QANet with PyTorch, using SQuAD 1.1.
Any contributions are welcome!
- Install pytorch 0.4 for Python 3.6+
- Run
pip install -r requirements.txt
to install python dependencies. - Run
download.sh
to download the dataset. - Run
python main.py --mode data
to build tensors from the raw dataset. - Run
python main.py --mode train
to train the model. After training,log/model.pt
will be generated. - Run
python main.py --mode test
to test an pretrained model. Default model file islog/model.pt
preproc.py: downloads dataset and builds input tensors.
main.py: program entry; functions about training and testing.
models.py: QANet structure.
config.py: configurations.
- The paper doesn't mention which activation function they used. I use relu.
- I don't set the embedding of
<UNK>
trainable. - The connector between embedding layers and embedding encoders may be different from the implementation of Google, since the description in the paper is inconsistent (residual block can't be used because the dimensions of input and output are different) and they don't say how they implemented it.
- Reduce memory usage
- Performance analysis
- Reach state-of-art scroes of the original paper
- Test on SQuAD 2.0
- Ablation analysis
- InitialBug: found two bugs: (1). positional encodings require gradients; (2). wrong weight sharing among encoders.
- linthieda: fixed one issue about dependencies and offered computing resources.