-
Notifications
You must be signed in to change notification settings - Fork 57
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Feat wget ckpt #40
Feat wget ckpt #40
Conversation
c71844e
to
3b94a44
Compare
src/zeroband/train.py
Outdated
if config.ckpt.resume is not None: | ||
# all is inplace | ||
ckpt_manager.load(resume_ckpt_path=config.resume) | ||
ckpt_manager.load(resume_ckpt_path=config.ckpt.resume) | ||
|
||
model.train() | ||
if elastic_device_mesh.live_recovery.need_live_recovery: | ||
ckpt_manager.download_and_load_ckpt_from_peers( | ||
elastic_device_mesh.live_recovery.get_adress(world_info.global_rank - 1) | ||
) | ||
elastic_device_mesh.live_recovery.need_live_recovery = False | ||
diloco.fake_step(model) |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do you need the load from resume_ckpt_path if you are doing live recovery? does it skip all the state live recovered state dict inits and only does dataloader loading?
add env var port live reco
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
NOICE! lgtm
live reco workig