Scaling laws for neural language models
OpenAI published the scaling laws paper establishing empirical power-law relationships between compute, data, and language model performance, fundamentally shaping LLM training.
Read at source: https://openai.com/index/scaling-laws-for-neural-language-models