Allennlp Optuna

allennlp-optuna: Hyperparameter Optimization Library for AllenNLP

`allennlp-optuna` is [AllenNLP](https://github.com/allenai/allennlp) plugin for hyperparameter optimization using [Optuna](https://github.com/optuna/optuna).

Supported environments

Machine \ Device	Single GPU	Multi GPUs
Single Node	:white_check_mark:	Partial
Multi Nodes	:white_check_mark:	Partial

AllenNLP provides a way of distributed training (https://medium.com/ai2-blog/c4d7c17eb6d6). Unfortunately, `allennlp-optuna` doesn't fully support this feature. With multiple GPUs, you can run hyperparameter optimization. But you cannot enable a pruning feature. (For more detail, please see [himkt/allennlp-optuna#20](https://github.com/himkt/allennlp-optuna/issues/20) and [optuna/optuna#1990](https://github.com/optuna/optuna/issues/1990))

Alternatively, `allennlp-optuna` supports distributed optimization with multiple machines. Please read the [tutorial](https://allennlp-optuna.readthedocs.io/en/latest/tutorial/hyperparameter_optimization_at_scale.html) about distributed optimization in `allennlp-optuna`. You can also learn about a mechanism of Optuna in the [paper](https://arxiv.org/pdf/1907.10902.pdf) or [documentation](https://optuna.readthedocs.io/en/stable/).

Documentation

You can read the documentation on [readthedocs](https://allennlp-optuna.readthedocs.io/).

1. Installation

pip install allennlp_optuna

# Create .allennlp_plugins at the top of your repository or $HOME/.allennlp/plugins
# For more information, please see https://github.com/allenai/allennlp#plugins
echo 'allennlp_optuna' >> .allennlp_plugins

2. Optimization

2.1. AllenNLP config

Model configuration written in Jsonnet.

You have to replace values of hyperparameters with jsonnet function `std.extVar`. Remember casting external variables to desired types by `std.parseInt`, `std.parseJson`.

``