Skip to content

yaml配置问题 #44

Description

@elonyu78

请问作者,按照官方设置per_device_train_batch_size为16,显存显然不够。我把batch size改为2,梯度累加设为4,分别率改为262144,sft得到的模型性能和论文相比还是有不小的差距(pretrain 模型是用你提供的)。请问您论文中最佳结果是用什么配置参数训练的,train loss最终能降到多少?

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions