EnochianJS

TS/JS library for programming LLM interactions. Read the docs here!

File Structure

docs/
js/
    src/
    examples/
studio/
sglang/

The docs/ directory contains mintlify docs. The js/ directory contains the enochian-js library. You can find the source code in js/src/ and examples of how to use it in js/examples/. The studio/ directory contains the enochian-studio library. The sglang/ directory is a submodule with the latest version of sglang Enochian is tested against.

Usage

To get started you will either need an OpenAI server or an SGLang server. You can an SGLang server on port 30000 like this:

python -m sglang.launch_server --model-path meta-llama/Meta-Llama-3-8B-Instruct --port 30000

Then install Enochian.

pnpm i enochian-js enochian-studio

To use Enochian, first initialize a ProgramState.

const s = new ProgramState()

Then, you can use normal JS control flow to "program" your LLM.

async function multiTurnQuestion(s: ProgramState, question1: string, question2: string): Promise<[string, string]> {
    await s.setModel('http://localhost:30000');
    await s.add(s.system`You are a helpful assistant.`)
        .add(s.user`${question1}`);
        .add(s.assistant`${s.gen('answer1')}`);
    await s.add(s.user`${question2}`)
        .add(s.assistant`No problem! ${s.gen('answer2')}`)
    return [s.get('answer1'), s.get('answer2')];
}

console.log(await multiTurnQuestion(s, 'Tell me a joke', 'Tell me a better one'));