Skip to content
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.

Sign up for GitHub

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

When save a model on TPU, make a copy to be moved to CPU #27993

Merged
merged 3 commits into from
Dec 19, 2023
Merged

When save a model on TPU, make a copy to be moved to CPU #27993

Changes from 2 commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
c658ffe
When save a model, make a copy to be moved to CPU, dont move the orig…
qihqi Dec 13, 2023
3cbee0f
make deepcopy inside of _save_tpu
qihqi Dec 13, 2023
13f4774
Move to tpu without copy
qihqi Dec 15, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
16 changes: 8 additions & 8 deletions src/transformers/trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2846,6 +2846,8 @@ def save_model(self, output_dir: Optional[str] = None, _internal_call: bool = Fa
def _save_tpu(self, output_dir: Optional[str] = None):
output_dir = output_dir if output_dir is not None else self.args.output_dir
logger.info(f"Saving model checkpoint to {output_dir}")
model = copy.deepcopy(self.model)
model.to("cpu")

if xm.is_master_ordinal():
os.makedirs(output_dir, exist_ok=True)
Expand All @@ -2854,22 +2856,20 @@ def _save_tpu(self, output_dir: Optional[str] = None):
# Save a trained model and configuration using `save_pretrained()`.
# They can then be reloaded using `from_pretrained()`
xm.rendezvous("saving_checkpoint")
if not isinstance(self.model, PreTrainedModel):
if isinstance(unwrap_model(self.model), PreTrainedModel):
unwrap_model(self.model).to("cpu").save_pretrained(
if not isinstance(model, PreTrainedModel):
if isinstance(unwrap_model(model), PreTrainedModel):
unwrap_model(model).save_pretrained(
output_dir,
is_main_process=self.args.should_save,
state_dict=self.model.state_dict(),
state_dict=model.state_dict(),
save_function=xm.save,
)
else:
logger.info("Trainer.model is not a `PreTrainedModel`, only saving its state dict.")
state_dict = self.model.state_dict().to("cpu")
state_dict = model.state_dict()
xm.save(state_dict, os.path.join(output_dir, WEIGHTS_NAME))
else:
self.model.to("cpu").save_pretrained(
output_dir, is_main_process=self.args.should_save, save_function=xm.save
)
model.save_pretrained(output_dir, is_main_process=self.args.should_save, save_function=xm.save)
if self.tokenizer is not None and self.args.should_save:
self.tokenizer.save_pretrained(output_dir)

Expand Down
Loading