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.

OnnxTextEmbedder

Runs an ONNX sentence-transformer and returns one vector per text.

public sealed class OnnxTextEmbedder : IDisposable

Constructor — takes the path to an ONNX model. Constructing it loads that model, which is why this type is the one place in Lodestar that needs a file you supply.

Properties — Dimension is the width of the vectors the model produces. MaxSequenceLength is the longest input, in tokens, the model accepts; longer inputs are truncated by the encoder rather than refused here.

Example — the shape of a call. It is not executed: see below.

using Lodestar.Embeddings.Onnx;

using var embedder = new OnnxTextEmbedder("model.onnx");

float[][] vectors = embedder.EmbedBatch(["a first sentence", "a second one"]);
int width = embedder.Dimension;

Remarks — every fence on this page and its members is docs-run: skip, and that is not an oversight. A running example would need a model of tens of megabytes, and weights are never committed to this repository — decisions/0003 is the rule, and the packaging sample declares the same exclusion for the same reason. The fences are still compiled against the packed package, so a renamed member still fails CI; only the values are unchecked, which is why none of them carries a // =>.

ONNX Runtime is referenced by this namespace and nowhere else in Lodestar. A consumer who never touches this type never pays for that dependency, which is the reason for the isolation.

It is IDisposable and holds native resources: the model session outlives garbage collection, so Dispose is not optional.

Applies to — net10.0, netstandard2.0.

See also — BatchEncoder, the embeddings guide, the Python equivalence table.

Members

Member	What it does
OnnxTextEmbedder.Dispose	Release the native model session.
OnnxTextEmbedder.Embed	One vector, from token ids you already have.
OnnxTextEmbedder.EmbedBatch	A vector per text, in one session run.