This year's baseline is a ensembled neural transition system based on the imitation learning paradigm introduced by Makarov & Clematide (2018).
-
The baseline requires Python 3.7. If your system does not use Python 3.7 by default (i.e., see
python --version
for your default Python), create and activate either:virtualenv --python=python3.7 sigmorphon source sigmorphon/bin/activate
- or a 3.7 Conda environment:
conda create --name=sigmorphon python=3.7 conda activate sigmorphon
-
Install the requirements and the library itself:
pip install --upgrade pip pip install -r requirements.txt pip install .
-
Run the sweep (may take a while):
./sweep
The baseline is made available under the Apache 2.0 license.
Makarov, P., and Clematide, S. 2018. Imitation learning for neural morphological string transduction. In Proceedings of the 2018 Conference on Empirical Methods in Natural Language Processing, pages 2877-2882.