Prerequisites

1. Installation was performed successfully. Refer to the Getting started: Neptune Cloud (preferred) or Getting started: local instance.
2. You are logged in to neptune

$ neptune login

3. In case you work locally (not is Neptune cloud) your virtualenv is activated.

$ source path/to/venv/toxic_venv/bin/activate

First experiment

Neptune cloud

$ neptune send --environment keras-2.0-gpu-py3 --worker gcp-gpu-medium --config neptune.yaml -- train_evaluate_predict_pipeline --pipeline_name glove_lstm

Note that data is already in the cloud and neptune.yaml is already filled with appropriate directories.

Locally

1. Download data to the directory defined in neptune.yaml#L29. It is data_dir
2. Run function that will perform train/validation split for you:

$ cd path/to/toxic-starter
$ python main.py train_valid_split

3. Start experiment

$ neptune run -- train_evaluate_predict_pipeline --pipeline_name glove_lstm