poon-llm

Connect to and stream from any OpenAI/Anthropic API. Lightweight, high performance, and simple, thoughtful API made for developers, encouraging use of CoT. Tested on OpenAI, Ollama, and Claude. Get better results from your LLM's using techniques that are made simple by poon-llm.

npm install poon-llm

OpenAI Example

import { OpenAI } from 'poon-llm';

const llm = new OpenAI({'secretKey': 'key', 'model': 'gpt-4o'});
const response = await llm.chat('Why is the sky blue?');

Ollama Example

const llm = new OpenAI({'apiBase': 'http://10.0.0.20', 'model': 'llama3'});
const response = await llm.chat('Why is the sky blue?');

Anthropic Example

import { Anthropic } from 'poon-llm';

const llm = new Anthropic({
    'secretKey': 'key',
    'model': 'claude-3-opus-20240229',
    'headers': {'Anthropic-Version': '2023-06-01'},
});
const response = await llm.chat('Why is the sky blue?');

Streaming

Streaming events occur at a fast rate, so to avoid crashing your server, poon-llm employs an efficient method to combat this: While an async onUpdate is executing, any chunks that come in will be ignored so that onUpdate will only be called as fast as your code can handle it. For example, if you are on a shared database that takes 1 second to write, your callbacks will fire back to back, after each write, and then once more at the very end.

const response = await llm.chat('Why is the sky blue?', {
    'onUpdate': text => Drafts.updateAsync({'_id': id}, {
        $set: {'body': text}
    }),
});

API Documentation

New Client - Options

Applies to new OpenAI(options), new Anthropic(options)

Name	Description
secretKey	Secret API key (Required for most API's)
apiBase	Specifies new base URL. Overrides the built-in defaults (Optional)
systemPrompt	Prompt to use for all chats
headers	Object containing headers to send (Required for Anthropic)

Chat Options

Applies to individual chat calls - llm.chat(message, options).

Option	Description
json	Enable JSON output: Requests underlying LLM API to respond in JSON, also JSON-parses and returns response. You must request the reply to be in JSON form in the system prompt. An error message will appear if the word JSON is not detected in the prompt.
xml	Array containing XML tags to parse. Causes the output to be an object with the keys specified by the array.
onUpdate	Callback function that is called every time the model has more chunks to append to the response.
context	Chat history for the conversation, must be an array of objects like {'role': String ('user' or 'assistant'), 'content': String}.
temperature	Float value controlling randomness in boltzmann sampling. Lower is less random, higher is more random.
maxTokens	Integer value controlling the maximum number of tokens generated.
prefill	String to prefill the LLM's response with. Useful for CoT.

Chain of Thought Example

Although JSON option is available, XML is generally better for prompts with Chain of Thought, because the LLM has an easier time formatting it, as it just needs to understand delimiters, rather than strict adherence to a certain syntax. XML is also easier to stream.

const response = await llm.chat(chatString, {
    'prefill': '<scratchpad>',
    'maxTokens': 2048,
    'xml': true,
});