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

Jump to bottom

added train_samples keyword for compliance check #645

Merged
merged 2 commits into from
May 30, 2024
Merged

added train_samples keyword for compliance check #645

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
7b97a6b
added train_samples keyword for compliance check
anmolgupt May 8, 2023
40e35b3
added cache clear logging
anmolgupt May 9, 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
9 changes: 9 additions & 0 deletions large_language_model/megatron-lm/megatron/training.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -95,6 +95,9 @@ def pretrain(train_valid_test_dataset_provider,
args_defaults: a dictionary from argument-name to argument-value. It
to set already parse arguments.
"""
# The reference implementation does not clear the cache currently
# but the submissions are required to do so
mllogger.event(key=mllogger.constants.CACHE_CLEAR, value=True)
mllogger.start(key=mllogger.constants.INIT_START, sync=False)

# Initalize and get arguments, timers, and Tensorboard writer.
Expand Down Expand Up @@ -232,6 +235,9 @@ def pretrain(train_valid_test_dataset_provider,
mllogger.event(key="trained_samples",
value=(args.consumed_train_samples - args.ext_lr_steps) * args.seq_length,
sync=False)
mllogger.event(key="train_samples",
value=(args.consumed_train_samples - args.ext_lr_steps) * args.seq_length,
sync=False)
mllogger.end(key=mllogger.constants.BLOCK_STOP,
metadata={'first_epoch_num': 0},
sync=False)
Expand Down Expand Up @@ -813,6 +819,9 @@ def train(forward_step_func, model, optimizer, opt_param_scheduler,
mllogger.event(key="trained_samples",
value=(args.consumed_train_samples - args.ext_lr_steps) * args.seq_length,
sync=False)
mllogger.event(key="train_samples",
value=(args.consumed_train_samples - args.ext_lr_steps) * args.seq_length,
sync=False)
if not saved_checkpoint:
save_checkpoint_and_time(iteration, model, optimizer,
opt_param_scheduler)
Expand Down
Loading