Skip to content

unsloth-studio

Recipe card from the charly-jupyter plugin (Images — the deployable catalog).

Unsloth Studio web UI for LLM fine-tuning with GPU acceleration.

Property Value
Base nvidia
Candies agent-forwarding, unsloth-studio, notebook-finetuning, dbus, charly
Platforms linux/amd64
Ports 8888, 8000
Registry ghcr.io/opencharly

The unsloth-studio candy is a Tier 2 environment-owner meta-layer that:

  1. Owns the pixi.toml (fine-tuning Python environment)
  2. Composes two Tier 1 sub-candies via candy: [llama-cpp, unsloth]
  3. Defines the supervisord service for the Studio web UI

Build order: pixi environment → llama-cpp (binaries) → unsloth (vLLM 0.19 wheel + unsloth pip + torch.compile patch) → supervisord config

  1. fedoranvidia (CUDA base)
  2. pixipythonsupervisord (transitive)
  3. unsloth-studio — Tier 2 meta-layer (owns pixi.toml, service config)
  4. llama-cpp — llama.cpp binaries (Tier 1, via candy:)
  5. unsloth — vLLM 0.19 + unsloth pip install + torch.compile patch (Tier 1, via candy:)
Port Service Protocol
8888 Unsloth Studio UI HTTP
8000 vLLM API server HTTP
Name Path Purpose
models ~/.cache/huggingface HuggingFace model cache
workspace /workspace Training data and outputs
Terminal window
charly box build unsloth-studio
charly config unsloth-studio
charly start unsloth-studio
# Open http://localhost:8888

After charly start:

  • charly status unsloth-studio — container running
  • charly service status unsloth-studio — all services RUNNING
  • curl -s -o /dev/null -w '%{http_code}' http://localhost:8888 — Studio HTTP returns 200

MUST be invoked when the task involves the unsloth-studio box, LLM fine-tuning via web UI, or Unsloth Studio deployment. Invoke this skill BEFORE reading source code or launching Explore agents.

  • /charly-image:image — image family umbrella (candy: image entries — those carrying base:/from: — in charly.yml, build/validate/inspect/list)
  • /charly-build:build — the embedded build vocabulary (distros, builders, init-systems)