neurovlm.training.MLPTextToBrainTrainConfig

neurovlm.training.MLPTextToBrainTrainConfig#

class neurovlm.training.MLPTextToBrainTrainConfig(output_root: 'str | Path' = 'runs', run_id: 'str | None' = None, seed: 'int' = 42, device: 'str' = 'auto', epochs: 'int' = 100, batch_size: 'int' = 256, eval_batch_size: 'int | None' = None, num_workers: 'int' = 0, learning_rate: 'float' = 5e-05, weight_decay: 'float' = 0.0, gradient_clip: 'float | None' = None, early_stopping_patience: 'int | None' = None, early_stopping_min_delta: 'float' = 0.0, max_train_batches: 'int | None' = None, max_eval_batches: 'int | None' = None, resume: 'str | Path | None' = None, preset: 'str' = 'retained', text_dim: 'int' = 768, hidden_dim: 'int' = 512, latent_dim: 'int' = 384, brain_dim: 'int' = 28542, primary_metric: 'str' = 'val_loss', metric_direction: 'MetricDirection' = <MetricDirection.MIN: 'min'>)[source]#
Parameters:

  • output_root (str | Path)

  • run_id (str | None)

  • seed (int)

  • device (str)

  • epochs (int)

  • batch_size (int)

  • eval_batch_size (int | None)

  • num_workers (int)

  • learning_rate (float)

  • weight_decay (float)

  • gradient_clip (float | None)

  • early_stopping_patience (int | None)

  • early_stopping_min_delta (float)

  • max_train_batches (int | None)

  • max_eval_batches (int | None)

  • resume (str | Path | None)

  • preset (str)

  • text_dim (int)

  • hidden_dim (int)

  • latent_dim (int)

  • brain_dim (int)

  • primary_metric (str)

  • metric_direction (MetricDirection)

__init__(output_root='runs', run_id=None, seed=42, device='auto', epochs=100, batch_size=256, eval_batch_size=None, num_workers=0, learning_rate=5e-05, weight_decay=0.0, gradient_clip=None, early_stopping_patience=None, early_stopping_min_delta=0.0, max_train_batches=None, max_eval_batches=None, resume=None, preset='retained', text_dim=768, hidden_dim=512, latent_dim=384, brain_dim=28542, primary_metric='val_loss', metric_direction=MetricDirection.MIN)#
Parameters:

  • output_root (str | Path)

  • run_id (str | None)

  • seed (int)

  • device (str)

  • epochs (int)

  • batch_size (int)

  • eval_batch_size (int | None)

  • num_workers (int)

  • learning_rate (float)

  • weight_decay (float)

  • gradient_clip (float | None)

  • early_stopping_patience (int | None)

  • early_stopping_min_delta (float)

  • max_train_batches (int | None)

  • max_eval_batches (int | None)

  • resume (str | Path | None)

  • preset (str)

  • text_dim (int)

  • hidden_dim (int)

  • latent_dim (int)

  • brain_dim (int)

  • primary_metric (str)

  • metric_direction (MetricDirection)

Return type:

None

Methods

__init__([output_root, run_id, seed, ...])

architecture()

Attributes

batch_size

brain_dim

device

early_stopping_min_delta

early_stopping_patience

epochs

eval_batch_size

gradient_clip

hidden_dim

latent_dim

learning_rate

max_eval_batches

max_train_batches

metric_direction

num_workers

output_root

preset

primary_metric

resume

run_id

seed

text_dim

weight_decay