About Songe

Songe is a continual learning engine for your local model. It runs alongside the model and keeps the traces, your instructions and every correction you make. When your agent is quiet, it trains the model on your own tasks, on your own machine.

A new version is used only if it passes checks. Nothing leaves your machine.

Needs

  • A model served locally by LM Studio, Ollama, llama.cpp, mlx-lm or vLLM
  • Training runs on Apple Silicon (MLX). Linux and Windows with NVIDIA or AMD are supported but less tested.
  • Machines without a GPU still serve and record.

Install

brew install sous-labs/tap/songe
songe init

Or pip install songe (Python 3.11 or later), then songe init.

songe init finds your model server and your agents, shows what it will change, and asks once. Then pick a songe/… model in your agent, or use any OpenAI-compatible tool.

Use

Nothing to do: use your agent. To look or change things:

songe          # what it is doing
songe open     # dashboard in your browser
songe models   # each model and its learning level
songe sleep    # learn now instead of waiting
songe undo MODEL   # go back one version
songe pause    # and: songe resume
songe forget   # delete recorded conversations

Reach out: admin@sous.pro