Skip to content

Update TFDS MLPerf data input pipeline#3747

Open
RissyRan wants to merge 1 commit intomainfrom
perf_input_pipeline
Open

Update TFDS MLPerf data input pipeline#3747
RissyRan wants to merge 1 commit intomainfrom
perf_input_pipeline

Conversation

@RissyRan
Copy link
Copy Markdown
Collaborator

@RissyRan RissyRan commented Apr 25, 2026

Description

Enable the dataset loading for MLPerf 6.0 DeepSeek V3.

Tests

Expect tests are green.

Local tests:

  • load new dataset: link
dataset_path=gs://mlperf-6-submission/tfds \
dataset_name='c4/en:3.0.5' \
eval_dataset_name='c4/en:3.0.5' \
  • load old dataset: link
dataset_path=gs://max-datasets-rogue \
dataset_name='c4/en:3.0.7' \
eval_dataset_name='c4/en:3.0.9' \

Checklist

Before submitting this PR, please make sure (put X in square brackets):

  • I have performed a self-review of my code. For an optional AI review, add the gemini-review label.
  • I have necessary comments in my code, particularly in hard-to-understand areas.
  • I have run end-to-end tests tests and provided workload links above if applicable.
  • I have made or will make corresponding changes to the doc if needed, including adding new documentation pages to the relevant Table of Contents (toctree directive) as explained in our documentation.

@codecov
Copy link
Copy Markdown

codecov Bot commented Apr 25, 2026

Codecov Report

❌ Patch coverage is 0% with 19 lines in your changes missing coverage. Please review.

Files with missing lines Patch % Lines
...t/input_pipeline/tfds_data_processing_c4_mlperf.py 0.00% 15 Missing ⚠️
src/maxtext/input_pipeline/input_pipeline_utils.py 0.00% 4 Missing ⚠️

📢 Thoughts on this report? Let us know!

Copy link
Copy Markdown
Collaborator

@shuningjin shuningjin left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Thanks, LGTM! Left some minor comments. The gpu test has been fixed at head; please sync.

else:
raise ValueError(f"{config.eval_dataset_name=} should be one of ('c4/en:3.0.1', 'c4/en:3.0.4', 'c4/en:3.0.5')")
raise ValueError(
f"{config.dataset_name=} should be one of "
Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

config.dataset_name -> config.eval_dataset_name

@@ -362,7 +385,11 @@ def make_c4_mlperf_eval_iterator(
is_tokenized_dataset = False
eval_slit = "validation"
Copy link
Copy Markdown
Collaborator

@shuningjin shuningjin Apr 26, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

nit: typo "eval_spit" -> "eval_split", there are four such instances inside function make_c4_mlperf_eval_iterator

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants