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

Remove skipping logic now that set_epoch exists #30501

Merged
merged 2 commits into from
Apr 26, 2024
Merged

Remove skipping logic now that set_epoch exists #30501

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
c05fc82
Remove skipping logic now that set_epoch exists
muellerzr Apr 26, 2024
d8b197d
Working version, clean
muellerzr Apr 26, 2024
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
19 changes: 0 additions & 19 deletions src/transformers/trainer.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -96,7 +96,6 @@
distributed_broadcast_scalars,
distributed_concat,
find_batch_size,
get_dataloader_sampler,
get_model_param_count,
get_module_class_from_name,
get_parameter_names,
Expand Down Expand Up @@ -2137,24 +2136,6 @@ def _inner_training_loop(

self.control = self.callback_handler.on_train_begin(args, self.state, self.control)

# Skip the first epochs_trained epochs to get the random state of the dataloader at the right point.
if not args.ignore_data_skip:
for epoch in range(epochs_trained):
sampler = get_dataloader_sampler(train_dataloader)
sampler_kinds = [RandomSampler]
if version.parse(accelerate_version) > version.parse("0.23.0"):
sampler_kinds.append(SeedableRandomSampler)
is_random_sampler = isinstance(sampler, tuple(sampler_kinds))
if not is_random_sampler:
# We just need to begin an iteration to create the randomization of the sampler.
for _ in train_dataloader:
break
else:
# Otherwise we need to call the whooooole sampler cause there is some random operation added
# AT THE VERY END!
sampler = sampler if sampler is not None else []
_ = list(sampler)

total_batched_samples = 0
for epoch in range(epochs_trained, num_train_epochs):
epoch_iterator = train_dataloader
Expand Down
Loading