• 2024-08-28Adam-mini Use Fewer Learning Rates To Gain More
    目录概Adam-mini代码ZhangY.,ChenC.,LiZ.,DingT.,WuC.,YeY.,LuoZ.andSunR.Adam-mini:Usefewerlearningratestogainmore.arXivpreprint,2024.概作者提出一种简化的optimizer,在取得和Adam相媲美的性能的同时,只需要一半的内存开销.Adam-mini