AI product Open source
Soup is an open-source Python CLI for fine-tuning and post-training large language models from a YAML configuration and a single training command. It supports supervised fine-tuning, LoRA, QLoRA, preference optimization, quantization, automatic batch-size and GPU detection, and local training without a cloud service.
Its optional layer-streaming mode keeps the frozen base model out of VRAM and feeds it to the GPU one decoder layer at a time, reducing memory use for low-VRAM systems. The project describes training an 8B model on a 4 GB laptop GPU with this mode; layer streaming is marked beta.
4 uses taken from transcripts — each links to the moment in the video.
An open-source Python CLI that simplifies LLM fine-tuning and post-training through YAML configuration and a single training command. It supports LoRA, QLoRA, supervised fine-tuning, preference optimization, and optional layer streaming to reduce VRAM use.
Fine-tunes LLMs locally from one YAML file and one command, handling batch size, GPU detection, quantization, and QLoRA. It also mitigates reward hacking during GRPO runs.
An open-source tool for fine-tuning language models from a YAML configuration and a single command. Its layer-streaming approach allows an 8B model to be fine-tuned on a laptop GPU with 4 GB of memory, and it supports DPO, knowledge editing, evaluation, signed adapters, and backdoor scanning.
An open-source tool for fine-tuning LLMs from a YAML file and a single command. Its layer-streaming approach allows an 8B model to be fine-tuned on a 4GB laptop GPU, with support for batching, quantization, model merging, and GGUF export.
4 in the library.