logitorch.pl_models.proofwriter

Classes

PLProofWriter

Initializes a PLProofWriter object.

Module Contents

class logitorch.pl_models.proofwriter.PLProofWriter(pretrained_model: str = 'google/t5-v1_1-large', learning_rate: float = None, weight_decay=0.1)

Bases: lightning.pytorch.LightningModule

Initializes a PLProofWriter object.

Args:

pretrained_model (str, optional): The name or path of the pretrained model to use. Defaults to “google/t5-v1_1-large”. learning_rate (float, optional): The learning rate for the optimizer. Defaults to None. weight_decay (float, optional): The weight decay for the optimizer. Defaults to 0.1.

configure_optimizers()

Configures the optimizer and scheduler for training.

Returns:

Tuple[List[Optimizer], List[Dict[str, Any]]]: The optimizer and scheduler.

forward(x, y) → transformers.modeling_outputs.SequenceClassifierOutput

Performs a forward pass of the model.

Args:

x: The input data. y: The target data.

Returns:

SequenceClassifierOutput: The output of the model.

predict(context: str, question: str, num_beams: int = 5, max_length: int = 120, device: str = 'cpu')

Generates predictions for the given context and question.

Args:

context (str): The context for the prediction. question (str): The question for the prediction. num_beams (int, optional): The number of beams for beam search decoding. Defaults to 5. max_length (int, optional): The maximum length of the generated sequence. Defaults to 120. device (str, optional): The device to use for prediction. Defaults to “cpu”.

Returns:

The generated predictions.

training_step(train_batch: Tuple[Dict[str, torch.Tensor], torch.Tensor], batch_idx: int) → torch.Tensor

Performs a training step.

Args:

train_batch (Tuple[Dict[str, torch.Tensor], torch.Tensor]): The batch of training data. batch_idx (int): The index of the batch.

Returns:

torch.Tensor: The loss value.

validation_step(val_batch: Tuple[Dict[str, torch.Tensor], torch.Tensor], batch_idx: int) → None

Performs a validation step.

Args:

val_batch (Tuple[Dict[str, torch.Tensor], torch.Tensor]): The batch of validation data. batch_idx (int): The index of the batch.

learning_rate = None

model

pretrained_model = 'google/t5-v1_1-large'

weight_decay = 0.1