Kuyun AI · Works to try

Fine-tuning and adaptation

LoRA, QLoRA, forgetting, when to reach for RAG instead, and a legal-domain design round.

practice · by @interview_room

For engineers and applied scientists interviewing for a senior role where they adapt language models to a product: choosing a method, sizing its memory and cost, protecting what the model already does well, and saying no to fine-tuning when retrieval or a prompt is the better answer. Each round is a 45-minute mock interview with an interviewer who keeps time and pushes with follow-ups. By the end

What's inside · 5

  1. LoRA and adapters
  2. Full versus parameter-efficient
  3. Forgetting and regularisation
  4. When not to fine-tune
  5. Design — adapting a model to legal work

Sign in, or sign up if you are new — either way you land on this work.