oneflow
oneflow copied to clipboard
fix the usage of argument end_factor in LinearLR
In LinearLR, when last_step >= total_iters,
lr = base_lr * 1.
to be consistent with pytorch
lr = base_lr * end_factor
related issue https://github.com/Oneflow-Inc/oneflow/issues/8343
Thanks for the fix in this pull request, it looks good to me. I will approve it which will be auto-merged to master.
The CI failed for code not formatted correctly.
Can you please run ninja of_format or make of_format for re-formating the code and then push it?
hi @basicv8vc
Thanks for the fix. I will merge it by a new PR https://github.com/Oneflow-Inc/oneflow/pull/9421.