📦 classifiers

A library for few-shot text classification, which is the task of classifying text into a set of categories, given only a few examples of each category.

Install

npm install @energetic-ai/classifiers

Usage

If you don't supply a specific model source, initClassifier will implicitly use remoteModelSource to download the model weights from TFHub on first use, which can take ~2 seconds.

import { initClassifier } from "@energetic-ai/classifiers";

(async () => {
  // Initialize with training examples
  const classifier = await initClassifier([
    ["The world is happy", "Positive"],
    ["Work is so fun", "Positive"],
    ["I had a great day", "Positive"],
    ["I had a bad day", "Negative"],
    ["I am frustrated", "Negative"],
    ["I am depressed", "Negative"],
  ]);

  // Classify a single string
  const single = await classifier.classify("The weather is so nice today");
  // { "label": "Positive" ... }

  // Classify multiple strings in a batch
  const multiple = await classifier.classify([
    "What is this? I am so angry!",
    "I am so excited!",
  ]);
  // [{ "label": "Negative" ... }, { "label": "Positive" ... }]
})();

You can also download the model weights locally by installing the @energetic-ai/model-embeddings-en package, and reduce cold start inference to ~20 ms at the cost of module size.

import { initClassifier } from "@energetic-ai/classifiers";
import { modelSource } from "@energetic-ai/model-embeddings-en";
(async () => {
  const model = await initClassifier(
    [
      /* ...examples... */
    ],
    modelSource
  );
  // ... snip ...
})();

Exports

Here's the table for the provided code:

Export	Description
ClassifierExample	A type representing an example for classification, consisting of two strings: input and label.
ClassifierExampleEmbedding	A type representing an example embedding for classification, consisting of an array of numbers (embedding) and a string (label).
ClassifierResult	A type representing the result of classification, containing the predicted label, class index, and confidence scores for each label.
Classifier	A class representing a classifier, which uses an embeddings model and a KNN classifier for classification.
Classifier.model	A property of the Classifier class representing the embeddings model used for classification.
Classifier.knn	A property of the Classifier class representing the KNN classifier used for classification.
Classifier.constructor	The constructor method of the Classifier class, used for initializing the classifier with examples and an embeddings model.
Classifier.classify(inputs)	A method of the Classifier class that classifies a single input string and returns a promise of ClassifierResult.
Classifier.classify(inputs[])	A method of the Classifier class that classifies an array of input strings and returns a promise of an array of ClassifierResult.
initClassifier	A function that initializes a classifier with examples and an optional embeddings model source.