PavelElk / yandex-algorithm-ml-track-2018

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Yandex Algorithm 2018 ML track

Final results: 2nd place

To reproduce the second place solution run blend.py script. It will blend predictions from 100 models.

Language and classification model training as well as prediction is in alica_fwd.ipynb file. Backwards model training and prediction (short version based on forward model) is in alica_bkw.ipynb file.

In order to run the solution install fast.ai library. Then replace two files (lm_rnn.py and text.py) from fastai folder.

About


Languages

Language:Jupyter Notebook 80.6%Language:Python 19.4%