Development build. This page describes main, not a released package. The latest published Lodestar.Embeddings is 0.8.0 — read its documentation.

Home › Embeddings › Tokenization

WordPieceVocabulary.Equals

Value equality over the entries and the settings.

public bool Equals(WordPieceVocabulary other)

Parameters — other is the vocabulary to compare against.

Returns — bool, true when both hold the same entries with the same ids and the same settings.

Example — the same file loaded twice.

using System.Text;
using Lodestar.Embeddings.Persistence;
using Lodestar.Embeddings.Tokenization;

var bounds = new ArtifactLoadOptions();
byte[] file = Encoding.UTF8.GetBytes("[UNK]\ntoken\n##ize\ntext");

WordPieceVocabulary first = VocabTxtLoader.Load(
    new MemoryStream(file), bounds, unkToken: "[UNK]", continuationPrefix: "##", lowercase: true);
WordPieceVocabulary second = VocabTxtLoader.Load(
    new MemoryStream(file), bounds, unkToken: "[UNK]", continuationPrefix: "##", lowercase: true);

bool same = first.Equals(second);  // => True

Remarks — the dictionary is compared by content, which a record's synthesised equality would not do: it would compare the two dictionaries by reference and report two loads of the same file as different vocabularies.

Comparing vocabularies is how to check that a saved tokenizer still matches the model beside it, which is worth doing once at start-up rather than inferring from bad output.

Applies to — net10.0, netstandard2.0.

See also — WordPieceVocabulary.