Method
Supervised fine-tuning
Full or domain fine-tuning of models on labeled or instruction data (SFT).
54 cases from 32 companies apply Supervised fine-tuning. 10 were checked in an online A/B test; 20 report results in production.
Typical tasks: Evaluate models, Build an assistant over internal docs, Label data with LLMs, Rank search results and 12 more. Companies: Yandex, Netflix, Yelp, DoorDash, Pinterest, Spotify and 26 more. Often combined with: RAG, LLM-as-a-judge, Knowledge distillation, Synthetic data generation (LLM), Reward model and 123 more.








