Clone, install & run

The first LydianAI release is a distributed training PoC (FedAvg + FastAPI). It's built to run across real-world mixed machines — including legacy Pascal GPUs.

Repo

Server/coordinator + workers + CLI client. Designed for multi-machine training over Tailscale.

Open GitHub

Typical folders: server/, worker/, client/, common/

Two install paths: NEW vs LEGACY GPUs

Modern GPUs use a current PyTorch stack. Pascal GPUs (GTX 1080/1080 Ti) require pinned legacy wheels and often Python 3.10.

Read install guide

Fast path

1) Bring machines on the same network

Use Tailscale so the controller and workers can talk over stable 100.x addresses.

Tailscale setup

2) Pick NEW or LEGACY GPU setup

Modern GPUs use new wheels; Pascal (GTX 1080/1080 Ti) needs pinned legacy Torch + Python.

GPU setup guide

Run the full Quickstart

CLI workflow

Start training, monitor rounds, check status/results, and list workers from a single CLI.

CLI reference