GaLore： Memory Efficient LLM Training by Gradient Low Rank Projection
На канале DS Talks https://t.me/+fQ07VSVJ2V8yZGYy разбирали статью GaLore: Memory-Efficient LLM Training by Gradient Low-Rank Projection https://arxiv.org/abs/2403.03507 Слайды https://kolodezev.ru/galore.html
На канале DS Talks https://t.me/+fQ07VSVJ2V8yZGYy разбирали статью GaLore: Memory-Efficient LLM Training by Gradient Low-Rank Projection https://arxiv.org/abs/2403.03507 Слайды https://kolodezev.ru/galore.html