:mod:`biotransformers.lightning_utils.models` ============================================= .. py:module:: biotransformers.lightning_utils.models Module Contents --------------- Classes ~~~~~~~ .. autoapisummary:: biotransformers.lightning_utils.models.LightningModule .. class:: LightningModule(model, alphabet, lr: float, warmup_end_lr: float, warmup_updates: int = 10, warmup_init_lr: float = 1e-07) Bases: :py:obj:`pytorch_lightning.LightningModule` Create lightning model to use ddp .. method:: forward(self, x) .. method:: configure_optimizers(self) -> Tuple[List[torch.optim.Optimizer], List[Dict]] Configure the optimizer and learning rate scheduler. :returns: - list of optimizers. - list of lr schedulers. .. method:: cross_entropy_loss(self, logits, targets) .. method:: training_step(self, train_batch, batch_idx) .. method:: validation_step(self, val_batch, batch_idx) Log the loss and metrics for a batch. :param batch: batch input. :param batch_idx: index of the batch. .. method:: get_tensor_accuracy(self, logits: torch.Tensor, targets: torch.Tensor) -> Tuple[torch.Tensor, torch.Tensor] Calculate accuracy for multi-masking, summed over batch. :param logits: prediction from the model, shape = (batch, len_tokens, len_vocab) :param targets: ground truth, shape = (batch, len_tokens) :returns: accuracy value.