Skip to main content

Tutorial Intro

Let's discover Igora Protocol and learn how to use it in less than 5 minutes.

Igora stands for I-ntelligent A-gora. With Agora meaning "market" in Modern Greek, a central public space in ancient Greek city-states.

The goal of Igora is to provide everyone, everywhere with llms (large language model) that can communicate with each other.

IGORA PROTOCOL uses 4 major basic concepts : clients ask workers to do tasks or jobs. The market is the place where workers and clients communicate.

Courses

To discover IA and LLM take a look at Courses

Understanding Igora concepts

What is a LLM -> https://www.youtube.com/watch?v=osKyvYJ3PRM

A Worker runs a LLM

You have certainely heard about Chat-GPT by OpenAI. This a is big example of LLM. But other solutions exists, that allow us to use LLMS has opensource models and to develop applications based on this technology.

  • A worker is a machine (computer, server...) that serve the LLM. You can ask the worker to generate a response to a prompt.
  • A prompt is a text that the LLM is asked to respond to.
  • A response is the text generated by the LLM.
  • A systemPrompt is a prompt that guide the worker to tell him how to generate a response (Maths teacher, Botanic expert...).

A worker offer its services on a market

  • Igora protocol is based on Yjs that allow us to create a distributed database of Workers where clients can ask a worker to accomplish a task.

A client create a job

  • A client create a job and ask for its completion on the market.
  • The market (with the help of a broker) will give the job to a worker.
  • The worker will do the job and give a response to the client.

Getting Started

There are two ways to get started with Igora Protocol:

  • You can create a Igora stack in local market (a worker and a client on your own computer).
    • Avantages : your data stay on your computer,
    • Disadvantages : you can not use other llms or workers.
  • You can use decentralized Igora market. Allowing you to use Igora protocol on a decentralized network. With this way, market, worker(s) and client(s) are distributed on the network. You specify which machine is a worker and which machine is a client. Advantages: Use a large bunch of llm , earn money or tokens with your workers