Reading the repo
Reading the repo
Assembly instructions
Fine-tune LLMs from one YAML. Layer streaming trains an 8B model on a 4 GB laptop GPU.
Saved setup from 3 h ago
Will it work on macOS?
Soup works with any text-generation model on the
HuggingFace Hub — if it loads with
AutoModelForCausalLM, it works, zero config changes. Llama 3.x/4, Qwen 2.5/3, Gemma 3, Mistral,
Mixtral, DeepSeek R1/V3, Phi-4, and 100+ others ship as ready-made recipes (soup recipes list).
| VRAM | Max model (QLoRA 4-bit) | Example | |---|---|---| | 8 GB | ~7B | Llama-3.1-8B, Mistral-7B | | 16 GB | ~14B | Phi-4-14B, Qwen2.5-14B | | 24 GB | ~34B | CodeLlama-34B, Yi-1.5-34B | | 48 GB | ~70B | Llama-3.3-70B | | 80 GB+ | 70B+ (full) or MoE | Mixtral-8x22B, DeepSeek-V3 |
Full model + vision tables and the optional-extras matrix are in docs/models.md.
pip install "soup-cli[train]" # add [train] to fine-tune; bare `soup-cli` is the light CLI
pipx install soup-cli
uv tool install soup-cli # same idea, if you already use uv
pipx install "soup-cli[train]"
pipx install "soup-cli[all]"
pipx install "git+https://github.com/MakazhanAlpamys/Soup.git"
pip install soup-cli
pip install "soup-cli[train]"
pip install "soup-cli[all]"
pip install git+https://github.com/MakazhanAlpamys/Soup.git
pip install "soup-cli[ui]"
From the README, exactly as the maintainers wrote it.
A self-hosted app: Docker runs it on your computer and you open it in the browser. Needs Docker Desktop (or Docker Engine on Linux). Run docker compose commands inside the cloned repo folder.
docker pull ghcr.io/makazhanalpamys/soup:latest
From the README, exactly as the maintainers wrote it.
No reports for macOS yet.
Opens a chat with this manual already filled in, for macOS. It uses your own account; gitmanual sends nothing.