composer_efficientnetb0

composer.models.composer_efficientnetb0(num_classes=1000, drop_connect_rate=0.2)

Helper function to create a ComposerClassifier with an EfficientNet-b0 architecture.

See Rethinking Model Scaling for Convolutional Neural Networks

(Tan et al, 2019) for more details.

Parameters

• num_classes (int, optional) – The number of classes. Needed for classification tasks. Default: 1000.

• drop_connect_rate (float, optional) – Probability of dropping a sample within a block before identity connection. Default: 0.2.

Returns

ComposerModel – instance of ComposerClassifier with a EfficientNet-B0 model.

Example:

from composer.models import composer_efficientnetb0

model = composer_efficientnetb0()  # creates EfficientNet-b0 for image classification