Models

Train a model

Train a new ML model on a dataset, or retrain an existing one.

To train a new model, pass name and omit ml_model_id. To retrain, pass ml_model_id and omit name; the model keeps its name and gains a new version. dataset_id and target_column are always required, and every column must exist in the dataset. Training runs in the background; check its progress with get_ml_model.

Requires an API key with the write scope.

post/api/v1/ml-models

Request

  • Base URL: https://platform.labric.co
  • URL: https://platform.labric.co/api/v1/ml-models
  • Auth: HTTP bearer

Request body

descriptionstring nullable
task_type'tabular' | 'multimodal' | 'segmentation'
target_columnstring required
quality_preset'best_quality' | 'high_quality' | 'good_quality' | 'medium_quality'
dataset_idstring uuid required
feature_columnsstring[] nullable
image_columnsstring[] nullable
problem_type'binary' | 'multiclass' | 'regression' | 'semantic_segmentation'
ml_model_idstring uuid nullable
namestring nullable

Response

OK

idstring uuid required
namestring required
descriptionstring nullable
task_typestring required
target_columnstring nullable
feature_columnsstring[] nullable
image_columnsstring[] nullable
problem_typestring nullable
currently_activeboolean
statusstring nullable
version_numberinteger nullable
dataset_idstring uuid nullable
quality_presetstring nullable
error_messagestring nullable
training_started_atstring date-time nullable
training_completed_atstring date-time nullable
evaluation_metricsobject nullable

Changes

No recorded changes to this operation across all 1 revision of this API.