On this week's AI board

Soup: fine-tune LLMs from one YAML file

Python CLI and web UI for fine-tuning and post-training LLMs (SFT, DPO, GRPO and more) from a single config file.