Obtain and Use Pre-trained Models

The Wolfram Neural Net Repository is a public resource that hosts an expanding collection of trained and untrained neural network models suitable for immediate evaluation, training, visualization, transfer learning and more. This example demonstrates a few of its models.

Load an image classification model from the repository.

Use the classifier on an image.

Load an image colorization model and copy the evaluation function from its resource page.

Use the function to colorize a grayscale image.

Load a semantic segmentation model and copy its evaluation function from its resource page.

Use a function to segment an image.

Load a speech recognition model.

Record an audio sample and transcribe it.

Load a language model and use it for word completion.

Autocomplete a list of words.

Load a particular model from a family of subword embeddings and use it to create a subword tokenizer.

Tokenize a string.

Obtain information about the allowed parameter values for this family of models.