Embeddings 0.4.0 bpetokenizer trygetid

Lodestar.Embeddings 0.4.0. This page is frozen at that release. Read the current documentation for what main says now. A link to a decision or a migration page follows main, and leaves the archive.

BpeTokenizer.TryGetId

The id of a token, if the vocabulary holds it.

public bool TryGetId(string token, out int id)

Parameters — token is the exact vocabulary entry, in its byte-level spelling where the model is byte-level. id receives the token's id, or 0 when the lookup fails.

Returns — bool, true when found; id is the id then and 0 otherwise.

Example — an entry that exists, and the byte-level spelling of a leading space.

using Lodestar.Embeddings.Tokenization;

var vocab = new Dictionary<string, int>(StringComparer.Ordinal)
{
    ["Ġ"] = 0, ["t"] = 1, ["o"] = 2, ["k"] = 3, ["e"] = 4, ["n"] = 5,
    ["to"] = 6, ["ken"] = 7, ["token"] = 8, ["Ġtoken"] = 9, ["ke"] = 10,
};
var merges = new List<MergePair> { new("t", "o"), new("k", "e"), new("ke", "n") };
var model = new BpeVocabulary(vocab, merges)
{
    ByteLevel = true,
    PreTokenizerPattern = BpePatterns.Gpt2,
    PreSplit = null,
};
var tokenizer = new BpeTokenizer(model);

bool plain = tokenizer.TryGetId("token", out int id);  // => True
bool spaced = tokenizer.TryGetId("Ġtoken", out int spacedId);  // => True
bool bare = tokenizer.TryGetId(" token", out int missing);  // => False

Remarks — " token" with a real space is not found, while "Ġtoken" is. In a byte-level model the space byte is represented by Ġ (U+0120), so that is how the entry is spelled and how it must be looked up. Searching a vocabulary with a literal space is the most common reason a token "is missing" when it is not.

token is found here although Encode never produces it — a lookup asks what the vocabulary holds, not what the merges can reach.

Applies to — net10.0, netstandard2.0.

See also — ISubwordTokenizer.TryGetId, BpeVocabulary.