Deploy from the SaladCloud Portal.

Overview

This guide explains how to deploy and run Kuzco (inference.net), a decentralized GPU network built on the Solana blockchain, using SaladCloud. Kuzco manages powerful AI inference workloads, particularly for large language models (LLMs) such as Llama3 and Mistral. SaladCloud simplifies deployment, allowing to easily provide compute resources to Kuzco and earn tokens ($INT points).

How to Use This Recipe

Step 1: Register an Account at Inference.net (Kuzco)

  1. Visit https://devnet.inference.net/register.
  2. Complete the registration form.
  3. Verify your email by clicking the confirmation link.

Step 2: Create Your Worker

  1. Log into your Kuzco dashboard.
  2. Navigate to the Workers tab.
  3. Skip to Step 3 if you already have a worker created.
  4. Click Create Worker at the top-right.
  5. Enter a descriptive worker name.
  6. Select Docker as the deployment type.
  7. Click Create Worker.

Step 3: Launch Your Worker

  1. You will be redirected to the Worker Details page.
  2. Click Launch Worker to generate and pull your unique worker code.

Step 4: Deploy Kuzco on SaladCloud

  1. Log into the Salad portal: https://portal.salad.com/
  2. Open an existing organization or create a new one.
  3. Click “Deploy a Container Group”.
  4. Scroll down to the Recipes section, find Kuzco (inference.net) and click “Get Started”.
  5. Enter a unique Container Group Name.
  6. Paste the Kuzco code you generated in the previous step.
  7. (Optional) Set the number of replicas depending on how many instances of a worker you want to run.
  8. (Optional) Click “Modify Configuration” to adjust advanced settings like GPU.
  9. Click Deploy.
SaladCloud will find a qualified GPU node and begin downloading the container image. This may take several minutes depending on the network and system conditions of the selected node. Once deployed, the container group will enter the Running state and show a green checkmark in the “Ready” column. Your Kuzco worker is now active. You can confirm the worker is live and earning by visiting the Worker Details page in your Kuzco dashboard: https://devnet.inference.net/dashboard/workers

Source and Documentation