npm package discovery and stats viewer.

Discover Tips

  • General search

    [free text search, go nuts!]

  • Package details

    pkg:[package-name]

  • User packages

    @[username]

Sponsor

Optimize Toolset

I’ve always been into building performant and accessible sites, but lately I’ve been taking it extremely seriously. So much so that I’ve been building a tool to help me optimize and monitor the sites that I build to make sure that I’m making an attempt to offer the best experience to those who visit them. If you’re into performant, accessible and SEO friendly sites, you might like it too! You can check it out at Optimize Toolset.

About

Hi, 👋, I’m Ryan Hefner  and I built this site for me, and you! The goal of this site was to provide an easy way for me to check the stats on my npm packages, both for prioritizing issues and updates, and to give me a little kick in the pants to keep up on stuff.

As I was building it, I realized that I was actually using the tool to build the tool, and figured I might as well put this out there and hopefully others will find it to be a fast and useful way to search and browse npm packages as I have.

If you’re interested in other things I’m working on, follow me on Twitter or check out the open source projects I’ve been publishing on GitHub.

I am also working on a Twitter bot for this site to tweet the most popular, newest, random packages from npm. Please follow that account now and it will start sending out packages soon–ish.

Open Software & Tools

This site wouldn’t be possible without the immense generosity and tireless efforts from the people who make contributions to the world and share their work via open source initiatives. Thank you 🙏

© 2026 – Pkg Stats / Ryan Hefner

forch

v1.0.0

Published

Framework para entrenar IAs conversacionales reales en Node.js. Aprende de miles de conversaciones humanas y genera respuestas contextuales coherentes.

Readme

forch

Framework de entrenamiento para IAs conversacionales en Node.js.

Entrena modelos de lenguaje reales (GPT-style) usando JavaScript puro. Sin dependencias de Python, sin configuraciones complejas. Solo npm install y a entrenar.

¿Qué hace?

Te permite entrenar un modelo desde cero con tus propios datos. No es un wrapper de OpenAI ni una API externa. Es tu propia red neuronal corriendo localmente.

  • Arquitectura real: Transformer decoder-only (como GPT-2/3) con atención causal.
  • Eficiente: Implementa KV-caching, gradient accumulation y optimizadores modernos (AdamW).
  • Conversacional: Diseñado específicamente para mantener el contexto en diálogos multi-turno.

Instalación

npm install forch

Uso Rápido

1. Entrenar

const { Agent, ConversationTrainer, BPE, ConversationDataset } = require('forch');

// 1. Tus datos
const conversations = require('./mis-datos.json');

// 2. Tokenizer
const tokenizer = new BPE();
tokenizer.train(conversations.flatMap(c => c.turns.map(t => t.content)), 1000);

// 3. Modelo
const agent = new Agent(tokenizer.nextId + 10, {
  embedDim: 128,
  numLayers: 4,
  numHeads: 4,
  maxLen: 512
});

// 4. Entrenar
const trainer = new ConversationTrainer({
  agent, 
  tokenizer, 
  dataset: new ConversationDataset(conversations),
  opts: { batchSize: 4, epochs: 10, lr: 5e-4 }
});

await trainer.train();

2. Conversar

const { ConversationManager } = require('forch');

const chat = new ConversationManager(agent, tokenizer);
const respuesta = chat.generateResponse('Hola, ¿cómo estás?');
console.log(respuesta);

Formato de Datos

Simplemente un JSON con tus conversaciones. Cuantos más ejemplos, mejor.

[
  {
    "turns": [
      { "role": "user", "content": "Hola" },
      { "role": "assistant", "content": "¡Hola! ¿En qué te ayudo?" }
    ]
  }
]

Rendimiento y Expectativas

Esto es JavaScript corriendo en CPU. No va a entrenar un modelo de 7B parámetros en 5 minutos.

  • Ideal para: Modelos pequeños/medianos (<100M params), prototipos, educación, y casos de uso específicos donde no quieres depender de Python.
  • Velocidad: Usa todos los núcleos disponibles, pero sigue siendo CPU.
  • Calidad: Con un buen dataset (1k+ conversaciones), los resultados son sorprendentemente buenos.

Scripts

  • npm run train: Corre el ejemplo de entrenamiento incluido.
  • npm test: Ejecuta la suite de pruebas.

Licencia

MIT. Úsalo para lo que quieras.

hecho con el corazon