arpes.deep_learning.interpret.Interpretation

class arpes.deep_learning.interpret.Interpretation(model, train_dataloader, val_dataloaders, train=True, val_index=0)

Provides utilities to interpret predictions of a model.

Importantly, this is not intended to provide any model introspection tools.

__init__(model, train_dataloader, val_dataloaders, train=True, val_index=0)

Methods

__init__(model, train_dataloader, ...[, ...])
dataloader_to_item_list(dataloader)	Converts data loader into a list of interpretation items corresponding to the data.
from_trainer(trainer)	Builds an interpreter from an instance of a pytorch_lightning.Trainer.
show([n_items, items, input_formatter, ...])	Plots a subset of the interpreted items.
top_losses(*[, ascending])	Orders the items by loss.

Attributes

items	All of the InterpretationItem instances inside this instance.
train
val_index
model
train_dataloader
val_dataloaders
train_items
val_item_lists