encode — Encode documents as sparse word or n-gram count matrices.

encode(bag, documents) encodes tokenizedDocument input or a row word vector against a RunMat bagOfWords or bagOfNgrams compatibility object.

Syntax

counts = encode(bag, documentsOrWords, Name, Value, ...)

Inputs

Returns

Errors

How encode works

• bag must be a scalar RunMat bagOfWords or bagOfNgrams object.
• documents can be a RunMat tokenizedDocument object. A non-tokenized row word vector is accepted as one document.
• The result is a RunMat sparse double matrix. By default, documents are rows and bag terms are columns.
• encode(___, 'DocumentsIn', 'rows') returns one row per document. This is the default.
• encode(___, 'DocumentsIn', 'columns') returns one column per document.
• encode(___, 'ForceCellOutput', true) wraps the sparse result in a 1-by-1 cell array. This preserves the documented scalar-bag cell-output form without pretending object-array parity exists.
• Words or n-grams not present in the bag model are ignored.
• For bagOfNgrams, RunMat encodes the n-gram rows stored on the compatibility object and derives scan lengths from those rows.
• Non-scalar bag object arrays and native MATLAB sparse model storage remain broader Text Analytics/runtime work.
• Integer and numeric model/document values are never converted to text. RunMat mode optionally accepts exact scalar integer ForceCellOutput predicates behind an extension gate; zero is false and every nonzero integer is true.

GPU memory and residency

encode returns host sparse output. It does not gather model or document arguments speculatively; invalid numeric resident values reject before provider access. A resident ForceCellOutput scalar is a separately gated RunMat extension.

Examples

Encode Documents With A Bag Of Words

bag = bagOfWords(tokenizedDocument(["alpha beta"; "beta gamma"]));
newDocuments = tokenizedDocument(["beta beta delta"; "alpha gamma"]);
counts = encode(bag, newDocuments)

Expected output:

`counts` is a sparse 2-by-3 matrix using `bag.Vocabulary` as the column order.

Put Documents In Columns

counts = encode(bag, ["alpha" "beta" "alpha"], "DocumentsIn", "columns")

Expected output:

`counts` is a sparse matrix with one column because the word vector is one document.

Encode N-Grams

bag = bagOfNgrams(tokenizedDocument("a b a"), "NgramLengths", [1 2]);
counts = encode(bag, tokenizedDocument("a b a b"))

Expected output:

`counts` contains unigram and bigram counts for n-grams stored in `bag.Ngrams`.

Using encode with coding agents

Open a RunMat example with live inputs, then ask the agent to explain how encode changes the result.

FAQ

Related Strings functions

Open-source implementation

Unlike proprietary runtimes, every RunMat function is open-source. Read exactly how encode is executed, line by line, in Rust.

• View the source for encode in Rust on GitHub
• Learn how the RunMat runtime works
• Found a bug? Open an issue with a minimal reproduction.

About RunMat

RunMat is an open-source runtime that executes MATLAB-syntax code blazing on any GPU. It is licensed under the Apache 2.0 license.

• RunMat automatically optimizes your math for GPU execution on Apple, Nvidia, and AMD hardware. No code changes needed. Simulations that took hours now take minutes.
• Start running code in seconds. RunMat runs in the browser, on the desktop, or from the CLI. No license server, no IT ticket.

Getting started · Benchmarks · Pricing