nip.image_classification.pretrained_models

nip.image_classification.pretrained_models#

Classes for loading pretrained image models, to use their embeddings.

Models are loaded from a ‘hub’ somewhere (e.g. PyTorch Image Models (timm) or Hugging Face Transformers).

The models are then used to extract embeddings from images, which can be used to aid agents in image classification tasks.

Classes

PretrainedImageModel(hyper_params, settings)

Base class for pretrained image models.

Resnet18Cifar10PretrainedModel(hyper_params, ...)

Resnet18 model trained on CIFAR-10.

Resnet18PretrainedModel(hyper_params, settings)

Base class for pretrained ResNet models using PyTorch Image Models (timm).