Closed aopolin-lv closed 1 year ago
We recommend using at least 8/16 A100-80G GPUs to fine-tune our model, and the training time depends on the specific task which may take a week or more. We have not yet explored what the minimum data requirement would be if one were to use smaller data to reduce training costs.
We recommend using at least 8/16 A100-80G GPUs to fine-tune our model, and the training time depends on the specific task which may take a week or more. We have not yet explored what the minimum data requirement would be if one were to use smaller data to reduce training costs.
谢谢
作者你好,看到你们的工作之后感觉非常有意思。请问本工作的训练计算资源要求是什么样的?例如说训练过程需要什么级别的显卡,多少张,训练多久?