admin 管理员组

文章数量: 1086019

I'm using torch LightningModule trainer. I create trainer with:

 trainer  = pl.Trainer(max_epochs = 3)

Each train epoch has 511 steps (total = 1533) and each validation epoch has 127 steps.

I use the self.global_step in training_step func and log it (with wandb):

 wandb.log({"train_step": self.global_step})

Zoom:

1 As you can see, it seems that the train_step in wandb contains the training steps and the validation steps. why ?

2 How can I view just the training values (with the training steps) in wandb ?

本文标签： torchWhy the globalstep (training step) is no sync with the wandb plot stepsStack Overflow

版权声明：本文标题：torch - Why the global_step (training step) is no sync with the wandb plot steps? - Stack Overflow 内容由网友自发贡献，该文观点仅代表作者本人， 转载请联系作者并注明出处：http://roclinux.cn/p/1744050214a2524881.html， 本站仅提供信息存储空间服务，不拥有所有权，不承担相关法律责任。如发现本站有涉嫌抄袭侵权/违法违规的内容，一经查实，本站将立刻删除。