A function wrapper for building an optimizer. Args: model (nn.Module or iterable): model. optim_cfg (CfgNode): optimization confi

r"""Count number of parameters in a model. Args: model (nn.Module): network model. params: network model`s params. Examples::

List non-hidden items in a directory. Args: path (str): directory path. sort (bool): sort the items.

A function wrapper for building a learning rate scheduler. Args: optimizer (Optimizer): an Optimizer. optim_cfg (CfgNode): optimi

Computes the accuracy over the k top predictions for the specified values of k. Args: output (torch.Tensor): prediction matrix with s

Create dirname if it is missing.

Build transformation function. Args: cfg (CfgNode): config. is_train (bool, optional): for training (True) or test (False).

Chooses static padding if you have specified an image size, and dynamic padding otherwise. Static padding is necessary for ONNX exporting of model

Generic training loops.

Check if an element is available in a list. Args: requested (str): probe string. available (list): a list of available strings.

1x1 convolution

Calculate and round number of filters based on depth multiplier.

Check if the given path is a file. Args: fpath (str): file path. Returns: bool

Create one-hot tensor. We suggest using nn.functional.one_hot. Args: label (torch.Tensor): 1-D tensor. num_classes (int): nu

Calculates the output image size when using Conv2dSamePadding with a stride. Necessary for static padding. Thanks to mannatsingh for pointing

3x3 convolution with padding

3x3 convolution with padding

r"""Load checkpoint. ``UnicodeDecodeError`` can be well handled, which means python2-saved files can be read from python3. Args:

Validates model name.

3x3 convolution with padding

Map EfficientNet model name to parameter coefficients.

Exponential moving average of model parameters. Args: model (nn.Module): model being trained. ema_model (nn.Module): ema of the m

Initialize the weights/bias with identity transformation.