unsloth-studio-layer
Recipe card from the charly-jupyter plugin (Images — the deployable catalog).
unsloth-studio – Fine-tuning Studio (Tier 2 meta-layer)
Section titled “unsloth-studio – Fine-tuning Studio (Tier 2 meta-layer)”Candy Properties
Section titled “Candy Properties”| Property | Value |
|---|---|
| Dependencies | cuda, supervisord |
| Sub-candies | llama-cpp, unsloth |
| Ports | 8888 (Studio UI), 8000 (vLLM API) |
| Volumes | workspace -> /workspace |
| Service | unsloth-studio (supervisord) |
| Install files | charly.yml, pixi.toml |
Architecture: Tier 2 Environment-Owner Meta-Layer
Section titled “Architecture: Tier 2 Environment-Owner Meta-Layer”This candy owns the pixi.toml for the fine-tuning environment and composes two Tier 1 candies via candy: [llama-cpp, unsloth]. Build order: pixi environment → llama-cpp (binaries) → unsloth (vLLM wheel + unsloth pip + patch) → supervisord config.
Environment Variables
Section titled “Environment Variables”| Variable | Value |
|---|---|
NVIDIA_PYTHON_PROJECT |
~/.pixi |
LD_LIBRARY_PATH |
/usr/lib64:$HOME/llama.cpp |
Plus from sub-candies: LLAMA_CPP_PATH, UNSLOTH_SKIP_LLAMA_CPP_INSTALL, HF_HOME
Packages (pixi.toml)
Section titled “Packages (pixi.toml)”Fine-tuning focused ML stack: PyTorch (CUDA 13.0), xformers, transformers, accelerate, vLLM runtime deps, HuggingFace (datasets, tokenizers, sentencepiece), fine-tuning (peft, trl, bitsandbytes, liger-kernel), GGUF tools
Service
Section titled “Service”Runs pixi run start-studio which executes unsloth studio -H 0.0.0.0 -p 8888. The Studio launches its own vLLM API server on port 8000 for inference and synthetic data generation.
Used In Boxes
Section titled “Used In Boxes”Related Candies
Section titled “Related Candies”/charly-jupyter:llama-cpp— Sub-candy: llama.cpp binaries/charly-jupyter:unsloth— Sub-candy: vLLM + unsloth pip install + patch/charly-infrastructure:supervisord— Process manager dependency/charly-jupyter:jupyter-ml— Alternative: ML Jupyter with MCP (same Tier 1 sub-candies)/charly-languages:python-ml— Alternative: core ML without UI
When to Use This Skill
Section titled “When to Use This Skill”Use when the user asks about:
- Unsloth Studio web UI setup
- The unsloth-studio service or container
- Port 8888 or 8000 in the context of fine-tuning
- The two-tier layer architecture for ML layers
- Fine-tuning pixi environment ownership
Related
Section titled “Related”/charly-image:layer— candy authoring reference (charly.ymlschema, plan steps, service declarations)/charly-check:check— declarative testing (check:block,charly check box,charly check live)