WildTrain CLI Reference

Complete command-line interface reference for WildTrain.

Main Commands

train

Train a model.

wildtrain train <task> [OPTIONS]

Arguments: - task: Task type (classifier/detector)

Options: - -c, --config PATH: Config file (required) - --resume PATH: Resume from checkpoint - --dry-run: Dry run without training

Examples:

# Train classifier
wildtrain train classifier -c configs/classification/train.yaml

# Train detector
wildtrain train detector -c configs/detection/yolo.yaml

eval

Evaluate a trained model.

wildtrain eval <task> [OPTIONS]

Arguments: - task: Task type (classifier/detector)

Options: - -c, --config PATH: Config file (required) - --checkpoint PATH: Model checkpoint - --split TEXT: Dataset split (test/val)

Examples:

wildtrain eval classifier -c configs/classification/eval.yaml
wildtrain eval detector -c configs/detection/yolo_eval.yaml

register

Register model to MLflow registry.

wildtrain register <model_type> <config>

Arguments: - model_type: Model type (classifier/detector) - config: Registration config file

Example:

wildtrain register detector configs/registration/detector_registration.yaml

tune

Run hyperparameter tuning.

wildtrain tune <task> [OPTIONS]

Options: - -c, --config PATH: Config file - --n-trials INTEGER: Number of trials

Example:

wildtrain tune classifier -c configs/classification/sweep.yaml --n-trials 50

serve

Start inference server.

wildtrain serve [OPTIONS]

Options: - -c, --config PATH: Inference config - --port INTEGER: Server port - --workers INTEGER: Number of workers

---

For detailed script documentation, see WildTrain Scripts.