search

brightmart / albert_zh

A LITE BERT FOR SELF-SUPERVISED LEARNING OF LANGUAGE REPRESENTATIONS, 海量中文预训练ALBERT模型
https://arxiv.org/pdf/1909.11942.pdf
3.93k stars 754 forks source link

albert_chinese_large 报显存错误 #149

Open teng1996 opened 3 years ago

teng1996 commented 3 years ago

请问albert_chinese_large 只有64Ｍ，为什么还显存溢出呢。用bert，同样的batch_size 并没有问题啊？

belle9217 commented 2 years ago

我用base v100 32G都能OOM