Training

Train your model using the prepared data and the architecture/hyperparameters specified in your config file.

$> # Train a model form the project's current settings, give it a name
$> koogu-train my_project.config my_first_model

The training step will:

1. Construct the model as described in the [model] section of your config file

2. Perform training and validation based on training/hyperparameter settings in the [train] section of your config file

3. Save trained model in [paths].model/my_first_model with all necessary files for testing and deployment

👉 Models directory path (model) is set under [paths] section of your config file.

During training, you will see epoch-by-epoch progress as below:

Epoch 1/50 ████████████████████████ 245/245 - 12s - loss: 0.3421 - binary_accuracy: 0.8234 - val_loss: 0.2891 - val_binary_accuracy: 0.8756

---

Parameters

Positional arguments

• <CONFIG FILE>

  Path to config file.

• <MODEL NAME>

  An identifier (name) for the model to be trained.

Options

• --seed NUM

  Seed value (integer) for deterministic shuffling.

Logging

• --log LOGFILE

  If specified, logging will be written out to this file instead of the default.

  Default: PROJECT-LOGS-DIR/train.log

• --loglevel LEVEL

  Logging level. Choices: CRITICAL, ERROR, WARNING, INFO, DEBUG.

  Default: INFO