Make sharding_tolerance configurable #1058
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Doris26
requested review from
gobbleturk,
jonb377,
khatwanimohit,
bvandermoon and
vipannalla
as code owners
|
Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA). View this failed invocation of the CLA check for more information. For the most up to date status, view the checks section at the bottom of the pull request. |
Doris26
changed the title
gobbleturk
requested changes
Dec 4, 2024
MaxText/train.py
Outdated
| @@ -71,6 +71,7 @@ | |||
| Transformer = models.Transformer | |||
| EPS = 1e-8 | |||
| _DEFAULT_OCDBT_TARGET_DATA_FILE_SIZE = 2 * 1024**3 | |||
| _DEFAULT_TOLERANCE = 0.02 | |||
There was a problem hiding this comment.
can you move this to a field in base.yml named e.g. "sharded_tolerance" defaulting to 0.02
MaxText/train.py
Outdated
| @@ -82,6 +83,8 @@ def validate_train_config(config): | |||
| if not config.base_output_directory.startswith("gs://"): | |||
| max_logging.log("WARNING: 'base_output_directory' might be pointing your local file system") | |||
| assert config.steps > 0, "You must set steps or learning_rate_schedule_steps to a positive integer." | |||
| if "tolerance" in config.__dict__ and (config.tolerance > 1.0 or config.tolerance < 0.0): | |||
There was a problem hiding this comment.
Can you move this check to pyconfig.py? Ideally we keep train.py as small as possible
MaxText/train.py
Outdated
| @@ -550,14 +553,15 @@ def setup_train_loop(config): | |||
| record_goodput(recorder, config, recorder.record_tpu_init_end_time if recorder else None) | |||
| record_goodput(recorder, config, recorder.record_training_preparation_start_time if recorder else None) | |||
| data_iterator, eval_data_iterator = create_data_iterator(config, mesh) | |||
| tolerance = config.tolerance if "tolerance" in config.__dict__ else _DEFAULT_TOLERANCE | |||
There was a problem hiding this comment.
this shouldn't be necessary when tolerance is part of config
MaxText/train.py
Outdated
|
|
||
| state, _, state_mesh_shardings, data_iterator = max_utils.setup_training_state( | ||
| model, data_iterator, tx, config, init_rng, mesh, checkpoint_manager | ||
| ) | ||
|
|
||
| if not config.using_pipeline_parallelism: | ||
| # The vocab tensor(s) of shape [vocab, embed] (and transpose) are not sharded by stage | ||
| maxtext_utils.assert_params_sufficiently_sharded(state.params, mesh, tolerance=0.02) | ||
| maxtext_utils.assert_params_sufficiently_sharded(state.params, mesh, tolerance) |
There was a problem hiding this comment.
you can use config.sharded_tolerance instead of tolerance here
gobbleturk
approved these changes
Dec 5, 2024
Doris26
force-pushed
the
tolerance_configurable
branch
3 times, most recently
from
737a915 to
614fb0b
Compare
Doris26
force-pushed
the
tolerance_configurable
branch
from
614fb0b to
2e0ac9d
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Using pure (512 DCN) FSDP triggers MaxText error of "Number of unsharded parameters exceeds tolerance 2% of total parameters."
Make tolerance a configurable param to avoid future errors across certain machines setups.
Checklist
Before submitting this PR, please make sure (put X in square brackets):