`kosmos.ml.config.sl_train`¶

Classes¶

class SLTrainConfig¶

Supervised learning training configuration.

dataset¶

Supervised learning dataset.

Type:: SLDataset

train_split¶

Fraction of dataset for training. Test split is 1 - train_split.

Type:: float

batch_size¶

Number of samples per batch.

Type:: int

num_epochs¶

Number of training epochs.

Type:: int

model_config¶

Model configuration.

Type:: ModelConfig

optimizer_config¶

Optimizer configuration.

Type:: OptimizerConfig

lr_scheduler_config¶

Learning rate scheduler configuration. Defaults to None.

Type:: LearningRateSchedulerConfig | None

max_grad_norm¶

Maximum gradient norm. Defaults to 1.0.

Type:: float | None

loss_config¶

Loss function configuration.

Type:: LossConfig

class FLTrainConfig¶

Bases: SLTrainConfig

Federated learning training configuration.