openai-tokens-count

OpenAI tokens calculator, with function calls, images, and messages in one call

Token Estimation

This package provides a utility function to estimate the token count for OpenAI chat completions.

Installation

To install the package, run the following command:

npm install openai-tokens-count

Usage

Here's an example of how to use the estimateTokens function:

import { estimateTokens } from 'openai-tokens-count';
import OpenAI from "openai"; // for typings

const message: OpenAI.Chat.ChatCompletionCreateParamsNonStreaming = {
  model: 'gpt-4-turbo',
  messages: [{ role: 'user', content: 'Hello' }],
};

const estimatedTokens = estimateTokens(message);
console.log('Estimated tokens:', estimatedTokens);

The function returns the estimated token count for the given input.

Testing

To run the tests, use the following command:

npm test

The tests are defined in the test/run.spec.ts file. They use the Jest testing framework to run test cases and compare the estimated token counts with the expected values.

Preparing Tests

To prepare test cases and generate the tokens.json file, follow these steps:

1. Create a directory named test/cases and add test case files inside it. Each test case file should export an object with the following properties:

   • model (string): The name of the OpenAI model to use for chat completion.
   • messages (array): An array of message objects, each containing a role and content property.

   Example test case file (hello-world.ts):

   export default {
     model: 'gpt-4-turbo',
     messages: [{ role: 'user', content: 'Hello' }],
   };

2. Run the tests/prepare-from-openai.ts script to generate the tokens.json file:

   npm run makeTests

   This script reads all the test case files from the test/cases directory, sends them to the OpenAI API for chat completion, and saves the actual token counts in the test/tokens.json file.

3. The tokens.json file will be used by the tests to compare the estimated token counts with the actual values.

   Example tokens.json file:

   {
     "hello-world.ts": 8
   }

Now you're ready to run the tests and verify the token estimation functionality.

Contributing

Contributions are welcome! If you find any issues or have suggestions for improvements, please open an issue or submit a pull request.

License

This package is open-source and available under the MIT License.