Basic PyTorch
implementation of
Neural Arithmetic Logic Units
by Andrew Trask, Felix Hill, Scott Reed, Jack Rae, Chris Dyer and Phil Blunsom.
- 4.1 Simple Function Learning Tasks
- 4.3 Language to Number Translation Tasks
git clone https://github.com/kerryeon/nalu
or download
Python
>= 3.6PyTorch
>= 0.4num2words
from pip
python ./test/simple_function.py
python ./test/language_to_number.py
Language to Number | Layer Size | Learning Rate | Plot |
---|---|---|---|
LSTM (en) | 32 | 0.01 | |
NALU (en) | 32 | 0.01 |
Basic pytorch implementation from bharathgs : https://github.com/bharathgs/NALU
- More Languages
- English
- Korean
- Japanese
- Roman
- Mayan
- Shuffle Dataset
- Furthermore
- 4.2 MNIST Counting and Arithmetic Tasks
- 4.4 Program Evaluation
- 4.5 Track Time in a Grid-World Environment
- Simplify code
- Comments