norm

batch_norm(g, node, name_to_tensor, null_ref, inference_target, **kwargs)

Translate operator aten::batch_norm to NNEF.

Nnef inputs:. input: tensor mean: tensor variance: tensor offset: tensor scale: tensor epsilon: scalar

nnef op

output = offset + scale * (input - mean) / sqrt(variance + epsilon);

group_norm(g, node, name_to_tensor, inference_target, **kwargs)

Translate operators aten::group_norm to NNEF.

It is a special case of NNEF batch_normalization. with variance and mean being tensor

layer_norm(g, node, name_to_tensor, null_ref, **kwargs)

Map PyTorch: 'aten:layer_norm', 'aten:native_layer_norm' to NNEF.

norm(g, node, name_to_tensor, inference_target, **kwargs)

NOTE this is only the normed vector.