From e266bd364e406c516672c014aad50d94542b6009 Mon Sep 17 00:00:00 2001
From: Laurianti <laurianti@gmail.com>
Date: Wed, 21 Oct 2026 15:36:57 +0200
Subject: [PATCH] Docs: ChromaDotNet.Embeddings, the embeddings of the Python
 client

docs/embeddings.md starts with ChromaDotNet.Embeddings and a short example, run against the packages of nuget.org on Chroma 1.5.9,
with its limits and the SearchByText sample of ChromaDotNet.Local; the README links it next to ChromaDotNet.Local.
---
 README.md          |  2 ++
 docs/embeddings.md | 25 +++++++++++++++++++++++++
 2 files changed, 27 insertions(+)

diff --git a/README.md b/README.md
index 9dc3d3c..d20bc25 100644
--- a/README.md
+++ b/README.md
@@ -58,6 +58,8 @@ dotnet add package ChromaDotNet.Client
 
 To run Chroma on your machine without Docker: [ChromaDotNet.Local](https://www.nuget.org/packages/ChromaDotNet.Local).
 
+To compute the embeddings as the Python client does: [ChromaDotNet.Embeddings](https://www.nuget.org/packages/ChromaDotNet.Embeddings), see [Embeddings](https://github.com/ChromaDotNet/ChromaDB.Client/blob/main/docs/embeddings.md).
+
 ## Example
 
 The models, like `ChromaRecords`, `ChromaQuery` and `ChromaCollectionDefinition`, are in `ChromaDB.Client.Models`. `AddChromaClient` is in `ChromaDB.Client.DependencyInjection`.
diff --git a/docs/embeddings.md b/docs/embeddings.md
index fc2331a..5c970f9 100644
--- a/docs/embeddings.md
+++ b/docs/embeddings.md
@@ -1,5 +1,30 @@
 # Embeddings
 
+## The embeddings of the Python client
+
+[ChromaDotNet.Embeddings](https://www.nuget.org/packages/ChromaDotNet.Embeddings) computes the vectors of the default embedding function of Chroma, the one the Python client uses when a collection has no other: the model all-MiniLM-L6-v2, with 384 dimensions. So .NET and Python read the same collection with the same embeddings.
+
+```
+dotnet add package ChromaDotNet.Embeddings --prerelease
+```
+
+```csharp
+using ChromaDB.Client;
+using ChromaDB.Client.Models;
+using ChromaDotNet.Embeddings;
+
+using var embeddings = await ChromaDefaultEmbeddingGenerator.CreateAsync();
+var collection = await client.GetOrCreateCollectionAsync(new ChromaCollectionDefinition("fruits") { EmbeddingGenerator = embeddings });
+var fruits = client.GetCollectionClient(collection).WithEmbeddingGenerator(embeddings);
+
+await fruits.AddAsync(["apple", "lemon"], documents: ["A crunchy apple", "A sour lemon"]);
+var nearest = await fruits.QueryAsync("something sour", nResults: 1);
+```
+
+The first `CreateAsync` downloads the model into the cache of the Python client, or uses the one Python downloaded. The definition declares the function `default`, as the Python client does. The package is a preview, and runs in 64-bit processes only, not on Intel Macs: see its [repository](https://github.com/ChromaDotNet/ChromaDB.Embeddings).
+
+The [SearchByText](https://github.com/ChromaDotNet/ChromaDB.Local/tree/main/samples/SearchByText) sample of ChromaDotNet.Local stores and searches texts this way, on a Chroma started without Docker. [Open it in a Codespace](https://codespaces.new/ChromaDotNet/ChromaDB.Local?quickstart=1).
+
 ## Embeddings from a generator
 
 ```csharp