Embeddings 0.6.0 bpevocabulary gethashcode - CyrilB1531/lodestar GitHub Wiki

Lodestar.Embeddings 0.6.0. 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.

BpeVocabulary.GetHashCode

A hash consistent with that equality.

public int GetHashCode()

Returns — int, over the flags and the two collections' sizes rather than their contents.

Example — equal models hash alike.

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 same = new BpeVocabulary(vocab, merges)
{
    ByteLevel = true,
    PreTokenizerPattern = BpePatterns.Gpt2,
    PreSplit = null,
};

bool equal = model.Equals(same);  // => True
bool hashesAlike = model.GetHashCode() == same.GetHashCode();  // => True

Remarks — a real BPE model holds fifty thousand entries and as many merges, and hashing all of them on every call would make the type unusable as a key. The size and the flags are enough to separate the models anyone actually compares, and two that collide are separated by Equals, which reads everything.

Applies to — net10.0, netstandard2.0.

See also — BpeVocabulary.Equals.

⚠️ **GitHub.com Fallback** ⚠️