README

# Llama API
curl "https://api.llama.com/v1/chat/completions" `
-H "Content-Type: application/json" `
-H "Authorization: Bearer <your llama api key>" `
-d '{
  "model": "Llama-4-Maverick-17B-128E-Instruct-FP8",
  "messages": [
    { "role": "user", "content": "Hello Llama! Can you give me a quick intro?" }
  ]
}'

# GPU instances from Nebius:

What you will be getting:

### Hardware:
* 16 vCPU
* 200 GB of RAM
* 1 Nvidia H100 GPU (80 GB of VRAM)
* 1 TB SSD storage

### Installed on the instance:
Ubuntu 22.04
Nvidia driver: v550.144.03 (CUDA v12.4)
Python 3.10
Docker Engine v28.0.1


## How to get your GPU instance

DM me (cyril.k), I will send you the credentials for connection via SSH.

## When you got your credentials
* Open your terminal
* Run `ssh llamahack@<host>` (replace `<host>` with your instance IP)
* You may be be prompted to add `<host>` to the list of known hosts, type `yes`
* Enter your `password` from credentials you received
* You are in, start hacking!

## Important
Do not attempt to update or re-install Nvidia driver on your instance (your VM already has it installed)