glm2和glm分别需要多大的显存才能微调
ShiXiangXiang123 opened this issue · comments
在batch-size=1的情况下?
十来G就够了
Tuning LLMs with no tears💦; Sample Design Engineering (SDE) for more efficient downstream-tuning.
ShiXiangXiang123 opened this issue · comments
在batch-size=1的情况下?
十来G就够了