Method
LoRA fine-tuning
Parameter-efficient fine-tuning with LoRA, QLoRA, P-tuning adapters.
42 cases from 28 companies apply LoRA fine-tuning. 11 were checked in an online A/B test; 13 report results in production. Includes variants: QLoRA.
Typical tasks: Build an assistant over internal docs, Evaluate models, Rank search results, Label data with LLMs and 9 more. Companies: Avito, Yandex, Pinterest, LY Corporation, LinkedIn, Celsus and 22 more. Often combined with: RAG, Supervised fine-tuning, Knowledge distillation, Prompt engineering, LLM-as-a-judge and 101 more.














