Research · Hacker News ·

Knowledge Distillation of Black-Box Large Language Models (2024)

Knowledge Distillation of Black-Box Large Language Models (2024)

A 2024 arXiv paper presents a knowledge distillation approach for black-box large language models. The work studies how to transfer behavior from inaccessible models into smaller models using output observations rather than internal parameters.

Read the full story at Hacker News →