Soup

---

Soup turns the pain of LLM fine-tuning into a simple workflow. One config, one command, done.

pip install "soup-cli[train]"   # add [train] to fine-tune; bare `soup-cli` is the light CLI
soup init --template chat
soup train

Fine-tune an 8B model on a 4 GB laptop GPU. Layer streaming keeps the frozen base out of VRAM and feeds it to the GPU one decoder layer at a time. Measured on an RTX 3050 Laptop 4 GB: Llama-3.1-8B-Instruct + NF4 at 119.6 tok/s at 952 MHz, 3.32 GB peak (pre-repair) — bit-exact against a normal resident run, and reproduced independently on an H100 at 113.00 tok/s in the same 3.32 GB. (Both figures were measured on v0.72.2, before the v0.73.0 correctness repair that cost −4.8% at 32B. The post-repair run, on another RTX 3050 Laptop 4 GB (@umran666, #1103), reads 208.6 tok/s at 1935–1957 MHz in a 2.40 GB peak. That is 59% of its same-session GEMM ceiling against 68%: about 14% slower per unit clock, so the raw 1.74x gain is the clock, not a speedup. Large-layer streaming also changed the memory pattern between the two rows, so even the fraction is not like-for-like — the record.) Opt-in (stream_layers: true) and still BETA — how it works · all measurements · paper · check it yourself on a free Colab T4 (caps the process to 4 GB, then asserts a streamed model is bit-identical to a normal one)

Why Soup?

Training LLMs is still painful. Even experienced teams spend 30-50% of their time fighting infrastructure instead of improving models. Soup fixes that.

• Zero SSH. Never SSH into a broken GPU box again.
• One config. A simple YAML file is all you need.
• Auto everything. Batch size, GPU detection, quantization — handled.
• Works locally. Train on your own GPU with QLoRA. No cloud required.

What's New

v0.77.2 — a cut of main after v0.77.1: 159 pull requests, 153 of them from 30 people outside the maintainer. Most of it is fixes. It also adds features and two rules that can refuse what v0.77.1 accepted, so read the v0.77.2 release notes before upgrading a server or a task: sft config. Coming from 0.76.x, run soup doctor --config soup.yaml first and read the v0.77.0 release notes.

• soup profile --model <id> answers "will it fit?" without a soup.yaml. It prices the run from the model's config.json, shaped by --seq, --batch, --method, --lora-r, --optimizer and --gradient-checkpointing; on Apple Silicon it compares with Metal's working set. soup plan and soup apply take --minillm-on-policy.
• More training on Apple Silicon. task: preference with preference_loss: dpo, orpo or simpo runs the native MLX trainer, MLX GRPO trains on 4-bit mlx-community bases, and MLX Adafactor follows the configured learning rate and schedule.
• Two new DPO recipes, qwen3.5-0.8b-dpo and qwen3.5-27b-dpo.
• Closer VRAM estimates. The optimizer step is its own phase, a PRM step is charged the logits of one position, a sharded FSDP or ZeRO run is priced per card, a stream_loss_chunk step its last layer's recompute, and Unsloth's LoRA targets as its loader picks them.
• Fewer silent overwrites and empty results. More commands refuse an --output that names an input (soup data ingest, persona-mix, generate and others); soup ship --adapter refuses an adapter that changes nothing; soup distill-prompt, soup data evolve, gen-magpie and filter --perplexity fail instead of writing an empty or unfiltered file.
• Preference and RL fixes. data.max_length holds again for SimPO and KTO, PPO checkpoints its policy when asked (and no longer crashes under 4-bit and 8-bit), and soup merge loads the base commit the adapter was trained on.
• Text shown as typed. Dozens of commands print paths, names and messages with brackets and :name: text kept and terminal control characters removed.

Behaviour changes to know. soup serve --tool-auth-token must be 16 to 128 characters of A-Z, a-z, 0-9, _ and - (exit 2 otherwise). task: sft refuses reasoning_effort, train_on_eot, use_flash_attn and use_liger on the vision, audio, MLX, Unsloth, RAFT and pre-tokenised paths that never applied them. Scripts from soup completions need regenerating. soup data from-traces --signal thumbs_up, MLX Adafactor, soup merge and the VRAM pre-flight can give a different result than on v0.77.1.

Older highlights live on the GitHub Releases page.

Quick Start

1. Install

Soup is a command-line application, so the cleanest install gives it its own environment and puts soup on your PATH:

# Light core: CLI + config + data tools, no PyTorch
pipx install soup-cli
uv tool install soup-cli          # same idea, if you already use uv

# Add the training stack (torch, transformers, peft, trl, datasets, …)
pipx install "soup-cli[train]"

# Everything (train + serve + ui + data) in one shot
pipx install "soup-cli[all]"

# Or from GitHub (latest dev)
pipx install "git+https://github.com/MakazhanAlpamys/Soup.git"

Already inside a virtualenv, a Colab notebook, or a Docker image? Use pip directly, with the same names and extras:

pip install soup-cli
pip install "soup-cli[train]"
pip install "soup-cli[all]"
pip install git+https://github.com/MakazhanAlpamys/Soup.git

Use pip rather than pipx if you also want to import soup_cli from your own code, since pipx deliberately isolates the application from everything else.

The full extras table (fast, mlx, serve, eval, ui, vision, audio, …) lives in docs/models.md.

error: externally-managed-environment? That is PEP 668, not a Soup problem. Debian 12, Ubuntu 23.04 and later stop pip from writing into the system Python, because apt manages those files too. pipx and uv tool sidestep it by giving Soup its own environment, which is why they are listed first above. python3 -m venv .venv && source .venv/bin/activate then plain pip works just as well.

Double quotes, not single. "soup-cli[train]" is the only spelling that works in every shell — cmd.exe, PowerShell, bash and zsh. If you copied 'soup-cli[train]' from an older tutorial and pip rejected it, that is the reason: why, and the exact error.

soup init, soup data …, and the other data/inspection commands work on the light install. Fine-tuning (soup train) needs the [train] extra.

2. Create a config

soup init                       # interactive wizard
soup init --template chat       # or start from a template

Templates: chat, code, tool-calling, medical, reasoning, vision, kto, orpo, simpo, ipo, bco, rlhf, pretrain, moe, longcontext, embedding, audio.

3. Train, test, ship

soup train --config soup.yaml                 # LoRA, quantization, batching — all handled
soup chat  --model ./output                    # talk to your model
soup push  --model ./output --repo you/my-model

soup merge  --adapter ./output                              # merge LoRA into the base
soup export --model ./output --format gguf --quant q4_k_m   # GGUF for Ollama / llama.cpp

More export targets (ONNX, TensorRT, BitNet) and deployment options live in docs/serving-and-export.md.

Web UI

Prefer a browser? soup ui serves a local dashboard for experiments, training setup, live metrics, dataset exploration and model chat.

pip install "soup-cli[ui]"
soup ui
# Opens http://127.0.0.1:7860/?token=<token> and prints the same URL

Web UI documentation

Configuration

A complete soup.yaml:

base: meta-llama/Llama-3.1-8B-Instruct
task: sft
# backend: unsloth  # 2-5x faster, pip install "soup-cli[fast]"

data:
  train: ./data/train.jsonl
  format: alpaca
  val_split: 0.1

training:
  epochs: 3
  lr: 2e-5
  batch_size: auto
  lora:
    r: 64
    alpha: 16
  quantization: 4bit

output: ./output

config/schema.py is the single source of truth for every field. Advanced data, training, and PEFT options are documented under Documentation.

Unknown config keys are rejected since v0.75. A key no model declares — a typo like quantizaton, or a field that only exists on a newer Soup — used to validate clean and be discarded, so the run proceeded with the setting simply not applied. v0.74 reported it at load with the field you probably meant; from v0.75 the same config fails to load, so fix or remove the key rather than relying on it being ignored. See Unknown config keys.

Documentation

The full feature reference lives in docs/. Start here:

Guide	Covers
Training tasks & methods	SFT, DPO/GRPO/PPO/KTO/ORPO/SimPO/IPO/BCO, tool-calling, PRM, pre-training, distillation, classification, vision/audio/TTS, unlearning, RAFT/RA-DIT, loop-hardening detectors
PEFT, long context & efficiency	DoRA, LoRA+, rsLoRA, VeRA, OLoRA, NEFTune, PiSSA, ReLoRA, optimizer & PEFT zoo, LLaMA Pro, GaLore, YaRN/LongLoRA, packing, curriculum, auto-tuning
Performance & quantization	QAT, FP8, Quant Menu (I + II), KV-cache, NVFP4, save formats, Cut Cross-Entropy, gradient checkpointing, kernels, activation offloading, layer streaming, multi-GPU / DeepSpeed / FSDP
Data engineering	Formats, the Axolotl/LF-parity pipeline, data tools, synthetic generation & forge, quality scorecards, trace tooling, remote datasets, mixing, recipe DAGs
Evaluation & probes	Eval design/gate, eval-gated training, benchmarks, NLG metrics, calibration, Elo arena, diagnose, post-train X-ray probes, A/B, drift, tunability, soup advise
Serving & export	OpenAI-compatible server, batch inference, benchmarking, merge/export, Anthropic Messages endpoint, speculative decoding (train + measure your own draft), deploy autopilot, Web UI, Agent Forge
Adapters, registry & governance	Adapter lifecycle/management, model registry, Soup Cans, the data flywheel (soup loop), knowledge editing, steering, supply-chain controls (scan/sign/BOM/attest/audit/airgap)
Compliance & governance quickstart	HIPAA/SOC2/EU-AI-Act/SR-11-7 init templates, provenance (BOM/attest/repro-receipt), audit log, air-gap, model-card autogen (soup card), CI gate (soup ci init)
Backends, platform & ops	MLX/Unsloth backends, alternative hubs, HF Hub integration, autopilot, experiment tracking, plan/apply, env lockfiles, hardware-fit, completions, plugins, utility commands
Command reference	The full soup command list
Supported models & extras	Recommended model families, the VRAM size guide, the pip extras matrix

Data Formats

Alpaca, ShareGPT, ChatML, preference pairs (DPO / ORPO / SimPO / IPO / KTO), vision, audio, ASR, plaintext, embedding, RAFT and more — all auto-detected from JSONL, JSON, CSV, Parquet or TXT, so in most cases you point data.train at a file and nothing else changes. Schemas with a worked example per format, plus the data pipeline (remote URIs, streaming, sharding, interleaving, vocab expansion, document ingestion), are in docs/data.md, with native Sesame CSM audio training documented separately.

Common Commands

soup train  --config soup.yaml        # train (SFT/DPO/GRPO/PPO/KTO/ORPO/SimPO/IPO/...)
soup infer  --model ./output --input prompts.jsonl --output results.jsonl   # batch inference
soup chat   --model ./output          # interactive chat
soup serve  --model ./output          # OpenAI-compatible API server
soup ui                               # local browser dashboard
soup merge  --adapter ./output        # merge LoRA into the base model
soup export --model ./output --format gguf           # export for deployment
soup eval   benchmark --model ./output               # evaluate
soup data   inspect ./data/train.jsonl               # dataset stats
soup recipes list                     # 100+ ready-made model recipes
soup autopilot --model <id> --data d.jsonl --goal chat  # zero-config
soup doctor                           # check GPU / deps / environment

The complete command list is in docs/commands.md.

Supported Models

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.

Docker

Run Soup without installing CUDA or PyTorch locally (image published to GHCR on every release):

docker pull ghcr.io/makazhanalpamys/soup:latest
docker run --gpus all -v $(pwd):/workspace ghcr.io/makazhanalpamys/soup train --config soup.yaml --yes
docker compose up   # or build locally

Requirements

• Python 3.11, 3.12 or 3.13 (those are the versions CI tests; 3.14+ is not supported yet because the PyTorch stack has not been validated there)
• GPU with CUDA (recommended), Apple Silicon (MPS), or CPU (experimental — very slow)
• 8 GB+ VRAM for 7B models with QLoRA

All training tasks run on CPU for testing (quantization auto-disabled). Optional extras (train, all, fast, vision, qat, serve, serve-fast, ui, eval, deepspeed, liger, mlx, onnx, tensorrt, …) are listed in docs/models.md.

Troubleshooting

soup doctor    # GPU, system resources, dependencies, and version in one place

CUDA wheels, version mismatches: docs/backends-and-ops.md.

Development

git clone https://github.com/MakazhanAlpamys/Soup.git
cd Soup
pip install -e ".[dev]"

ruff check src/soup_cli/ tests/    # lint
pytest tests/ -v                   # unit tests (fast, no GPU)
pytest tests/ -m smoke -v          # smoke tests (downloads a tiny model, trains)
pytest tests/ -m gpu --no-cov -v   # GPU tests (need a CUDA card; report results, see CONTRIBUTING.md)

pre-commit install                 # optional: ruff lint+format on commit

See CONTRIBUTING.md for the full workflow and SECURITY.md to report a vulnerability. Telemetry is strictly opt-in (SOUP_TELEMETRY=1, default off; see Privacy Policy).

Support Soup

Soup is Apache-2.0 and free — and stays that way. It is built and maintained in the open on a single 4 GB laptop, which is why every performance number in these docs is measured rather than claimed.

If Soup saved you a training run, starring the repo helps most, and it costs nothing. If you would like to fund the work directly:

❤️ Donate — one-off, any amount (use Change amount on the checkout page). Payments are processed by Stripe under the maintainer's registered business, MePlay, Inc. — that name, not "Soup", is what appears on the checkout page and on your card statement.

Donations buy GPU time for the hardware-gated work — multi-GPU, 8B+ validation, Apple Silicon — that a single 4 GB laptop cannot reach.

The other way to move exactly those items is hardware itself. They ship behind honest "requires <hardware>" gates rather than unverified claims, so if you have access to a bigger box — or GPU credits going unused — running one of the help wanted issues and posting the numbers helps as much as funding the GPU time would. Those issues say exactly what is blocked on hardware today.

Contributors

Built by the community ❤️ — thank you to everyone who has contributed. See CONTRIBUTORS.md.

Contact

Bugs and feature requests belong in the issue tracker, questions in Discussions — both get answered faster and help the next person with the same problem.

For live chat, setup help, and everything that reads better as a conversation, join the Discord or the Telegram community. Anything that should still be findable in six months belongs in Issues or Discussions — a Discord answer helps one person, an issue helps everyone who hits the same thing. The Code of Conduct applies there too.

For anything that does not fit in public — security reports (see SECURITY.md), Code of Conduct matters, or press — email [email protected]. That is the project address and the right one for anything Soup-related. [email protected] is the maintainer's personal address; it reaches the same person and is a fine fallback.

Citing Soup

Layer streaming — training an 8B model on a 4 GB laptop GPU by streaming the frozen base from host RAM one decoder layer at a time — is described in a preprint, together with the correctness protocol that verifies a streamed run against a resident one (forward and backward stated separately, because they are two claims and not one).

Makazhan, A. (2026). Exact Layer Streaming: LoRA Fine-Tuning of an 8B Model on a 4 GB Laptop GPU (v3). Zenodo. https://doi.org/10.5281/zenodo.21918325

Version 3 (13 August 2026) is current. The title and the claim are unchanged — 8B on 4 GB — and no measured number has changed since v1. What v3 does is withdraw an explanation we had published, which is also the shortest way to describe what the paper is for:

• Retracted in v3: "layer streaming is bound by host-to-device transfer, not by the GPU." That was an inference from the H100 replication below, and it had never been measured. We measured it on 11 August and it is false at the published configuration: deleting every host-to-device byte buys 1.4%, the compute stream waits on a copy for 0.20% of the step, and the step runs at 71.3% of that card's same-session GEMM ceiling. The largest streaming-specific cost is the per-layer NF4 dequantisation, at 9.8% (the record). Every measurement stands; the replication survives in a weaker form — the constraint is common to both machines and is not the GPU's compute.
• Replication on hardware nothing like the original (added in v2): 119.6 tok/s on the RTX 3050 (at 952 MHz) against a median 113.00 on an H100, at the same 3.32 GB peak. Both pre-date the #331 repair and are the figures the paper reports. The post-#331 row (@umran666, #1103) is 208.6 tok/s at 1935–1957 MHz in a 2.40 GB peak, 59% against 68% of the same-session GEMM ceiling: the raw 1.74x gain is the clock, not a speedup, and the memory pattern changed between the rows, so even the fraction is not like-for-like (the record).
• A silent wrong-gradient defect, found and repaired. On NF4 above ~165 MiB per layer the forward stayed bit-exact and the loss curve looked healthy while the gradients were wrong. The cause is named in the upstream library and reported there; the repair is gated against controls on real 32B and 72B.
• Bit-exactness at real model sizes instead of three-layer toys: forward from 0.5B to 72B, backward at 8B and 14B.
• Trained-model quality, measured for the first time, and indistinguishable from a resident run.
• A comparison against DeepSpeed — including the result that does not flatter us: eight cards of ZeRO-3 are slower than one card training resident.
• The limitations section rewritten: of v1's ten items, one closed and four more narrowed, and seven new ones added.

Cite the version you used. 10.5281/zenodo.21771064 is the concept DOI and always resolves to the latest version (v3 today); v1 and v2 remain citable at their own version DOIs and are not edited — the retraction above is a new version precisely so that the record of what we claimed, and when, stays intact.

The measurement records behind every number in it are in benchmarks/, published as written — including the failures, the assumptions that turned out wrong, and the numbers that were measured and then discarded.

@misc{makazhan2026exact,
  title        = {Exact Layer Streaming: LoRA Fine-Tuning of an 8B Model on a 4 GB Laptop GPU},
  author       = {Makazhan, Alpamys},
  year         = {2026},
  publisher    = {Zenodo},
  version      = {v3},
  doi          = {10.5281/zenodo.21918325},
  url          = {https://doi.org/10.5281/zenodo.21918325}
}

License

Apache-2.0. Copyright 2026 Makazhan Alpamys and the Soup contributors. If you redistribute Soup or a modified copy of it, sections 4 and 6 of the license apply to you: keep LICENSE and NOTICE with every copy, mark the files you changed, and do not use the Soup name beyond saying where the work comes from.