Soup

Fine-tunes and post-trains LLMs from one YAML config, streaming large models onto small consumer GPUs.

Desktopfreeglobal

Soup is an open-source command-line tool for fine-tuning and post-training large language models from a single YAML configuration file. Its layer-streaming technique lets an 8-billion-parameter model train on a 4GB consumer laptop GPU by keeping the frozen base model out of VRAM, and it supports both supervised fine-tuning and preference-tuning methods. It's built for developers and ML practitioners who want to fine-tune models locally without managing cloud GPU infrastructure.

Website

Full match profile

Behind the summary, Matchbox keeps a richer profile of Soup - the signals our matcher actually reads to decide when to surface it. It stays private; claim the listing to see and control it.

Problem & pain-point mapping
Who we surface it to (audience fit)
What it's a strong alternative to
Trust & credibility signals

Claim Soup to unlock →

Try Matchbox with your own problem

Describe what is not working - we’ll show you whether Soup (or something else) actually fits.

Try a problem