Esc
Ask AIAnswers may be inaccurate; check the linked pages.Esc
Ask anything about these docs, like how to get started or what a function does.

LagLlamaLightningModule

lightning_module.LagLlamaLightningModule

A pl.LightningModule class that can be used to train a LagLlamaLightningModule with PyTorch Lightning. This is a thin layer around a (wrapped) LagLlamaLightningModule object, that exposes the methods to evaluate training and validation loss.

__init__(self, model_kwargs: dict, context_length: int, prediction_length: int, loss: DistributionLoss = NegativeLogLikelihood(), lr: float = 0.001, weight_decay: float = 1e-08, aug_prob: float = 0.1, freq_mask_rate: float = 0.1, freq_mixing_rate: float = 0.1, jitter_prob: float = 0.0, jitter_sigma: float = 0.03, scaling_prob: float = 0.0, scaling_sigma: float = 0.1, rotation_prob: float = 0.0, permutation_prob: float = 0.0, permutation_max_segments: int = 5, permutation_seg_mode: str = 'equal', magnitude_warp_prob: float = 0.0, magnitude_warp_sigma: float = 0.2, magnitude_warp_knot: int = 4, time_warp_prob: float = 0.0, time_warp_sigma: float = 0.2, time_warp_knot: int = 4, window_slice_prob: float = 0.0, window_slice_reduce_ratio: float = 0.9, window_warp_prob: float = 0.0, window_warp_window_ratio: float = 0.1, window_warp_scales: list = [0.5, 2.0], data_id_to_name_map: dict = {}, use_cosine_annealing_lr: bool = False, cosine_annealing_lr_args: dict = {}, track_loss_per_series: bool = False, nonnegative_pred_samples: bool = False, use_kv_cache: bool = True, use_single_pass_sampling: bool = False)

Parameter Type Default Description
model_kwargs dict - (undocumented)
context_length int - (undocumented)
prediction_length int - (undocumented)
loss DistributionLoss NegativeLogLikelihood() Loss function to be used for training.
lr float 1e-3 Learning rate.
weight_decay float 1e-8 Weight decay regularization parameter.
aug_prob float 0.1 (undocumented)
freq_mask_rate float 0.1 (undocumented)
freq_mixing_rate float 0.1 (undocumented)
jitter_prob float 0.0 (undocumented)
jitter_sigma float 0.03 (undocumented)
scaling_prob float 0.0 (undocumented)
scaling_sigma float 0.1 (undocumented)
rotation_prob float 0.0 (undocumented)
permutation_prob float 0.0 (undocumented)
permutation_max_segments int 5 (undocumented)
permutation_seg_mode str 'equal' (undocumented)
magnitude_warp_prob float 0.0 (undocumented)
magnitude_warp_sigma float 0.2 (undocumented)
magnitude_warp_knot int 4 (undocumented)
time_warp_prob float 0.0 (undocumented)
time_warp_sigma float 0.2 (undocumented)
time_warp_knot int 4 (undocumented)
window_slice_prob float 0.0 (undocumented)
window_slice_reduce_ratio float 0.9 (undocumented)
window_warp_prob float 0.0 (undocumented)
window_warp_window_ratio float 0.1 (undocumented)
window_warp_scales list [0.5, 2.0] (undocumented)
data_id_to_name_map dict {} (undocumented)
use_cosine_annealing_lr bool False (undocumented)
cosine_annealing_lr_args dict {} (undocumented)
track_loss_per_series bool False (undocumented)
nonnegative_pred_samples bool False (undocumented)
use_kv_cache bool True (undocumented)
use_single_pass_sampling bool False (undocumented)

forward(self, *args, **kwargs)

Performs greedy prediction/sampling.

Parameters:

Parameter Type Default Description
*args - - (undocumented)
**kwargs - - (undocumented)

training_step(self, batch, batch_idx: int)

Execute training step.

Parameters:

Parameter Type Default Description
batch - - (undocumented)
batch_idx int - (undocumented)

on_train_epoch_end(self)

(undocumented)

validation_step(self, batch, batch_idx: int)

Execute validation step.

Parameters:

Parameter Type Default Description
batch - - (undocumented)
batch_idx int - (undocumented)

on_validation_epoch_end(self)

(undocumented)

configure_optimizers(self)

Returns the optimizer to use.