liveBook · Manning

Multi-hot tensor

As explained in Natural Language Processing in Action, Second Edition

A multi-hot vector is a binary vector with any number of active positions indicating multiple selected labels from a fixed set.

A multi-hot tensor represents multiple selected labels with zeros and ones. Each position corresponds to a fixed label, and several positions—or none—may be active at once. section 9.3.4

Toxicity tags

For labels [toxic, severe, obscene, threat, insult, hate], [1, 0, 0, 0, 0, 0] marks a toxic comment, while [0, 0, 0, 0, 0, 0] marks a comment with no selected labels. A comment can activate several positions simultaneously. section 9.3.4

How Natural Language Processing in Action, Second Edition covers it

How deep it goes
Defines binary positions, zero-or-more active labels, and constant dimensions, then demonstrates dataset conversion without detailing tensor internals.
Examples
Uses toxic-comment labels and lists such as [1, 0, 0, 0, 0, 0] and [0, 0, 0, 0, 0, 0].
What it assumes
Assumes familiarity with labels and classification while clarifying why multi-label outputs differ from conventional one-hot targets.
What you can do after
Apply it via simpletransformers by converting label columns into fixed-length arrays for MultiLabelClassificationModel.
How it connects
Connects multi-label classification with one-hot vectors, multiclass classification, MultiOutputClassifier, and transformer models.
Cover of Natural Language Processing in Action, Second Edition
liveBookmanning.com
Read Natural Language Processing in Action, Second Edition on liveBook Get the book at manning.com

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Natural Language Processing in Action, Second Edition. The section links point to the parts of the book it draws on.