def evaluate_batch(self, batch: Any, model: nn.Module) -> Dict[str, Any]:
batch = cast(Tuple[TorchData, Dict[str, torch.Tensor]], batch)
data, labels = batch
output = model(data)
error = error_rate(output["digit_predictions"], labels["digit_labels"])
return {"validation_error": error}
def train_batch(self, batch: TorchData, model: nn.Module, epoch_idx: int,
batch_idx: int) -> Dict[str, torch.Tensor]:
batch = cast(Tuple[torch.Tensor, torch.Tensor], batch)
data, labels = batch
output = model(data)
loss = torch.nn.functional.nll_loss(output, labels)
error = error_rate(output, labels)
return {"loss": loss, "train_error": error}
def evaluate_batch(self, batch: TorchData,
model: nn.Module) -> Dict[str, Any]:
batch = cast(Tuple[torch.Tensor, torch.Tensor], batch)
data, labels = batch
output = model(data)
error = error_rate(output, labels)
return {"validation_error": error}
def train_batch(self, batch: Any, model: nn.Module, epoch_idx: int,
batch_idx: int) -> Dict[str, torch.Tensor]:
batch = cast(Tuple[TorchData, Dict[str, torch.Tensor]], batch)
data, labels = batch
output = model(data)
loss = compute_loss(output, labels)
error = error_rate(output["digit_predictions"], labels["digit_labels"])
return {"loss": loss, "classification_error": error}
def evaluate_batch(self, batch: TorchData,
model: nn.Module) -> Dict[str, Any]:
"""
Calculate validation metrics for a batch and return them as a dictionary.
This method is not necessary if the user overwrites evaluate_full_dataset().
"""
batch = cast(Tuple[torch.Tensor, torch.Tensor], batch)
data, labels = batch
output = model(data)
error = error_rate(output, labels)
return {"validation_error": error}