[Doc] Add iter_jax_batches documentation to JaxTrainer guide#63294
Merged
ryanaoleary merged 8 commits intoMay 12, 2026
Conversation
Signed-off-by: Ryan O'Leary <ryanaoleary@google.com>
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request updates the JAX trainer tutorial documentation in both the notebook and Markdown formats to include an alternative section on using the iter_jax_batches API for native JAX data ingestion. The reviewer suggested renaming the local_batch variable to batch or global_batch to better reflect that the API yields globally sharded arrays. Additionally, it was recommended to explicitly label iter_jax_batches as an "Alpha API" in the Markdown file for consistency and accuracy.
Signed-off-by: Ryan O'Leary <ryanaoleary@google.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> Signed-off-by: Ryan O'Leary <113500783+ryanaoleary@users.noreply.github.com>
Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> Signed-off-by: Ryan O'Leary <113500783+ryanaoleary@users.noreply.github.com>
Signed-off-by: Ryan O'Leary <ryanaoleary@google.com>
Contributor
Author
|
cc: @siyuanfoundation @liulehui added a section on the |
liulehui
reviewed
May 12, 2026
|
|
||
| In the example above, we used `iter_batches(batch_format="numpy")` and manually transferred data to devices using `jax.make_array_from_process_local_data`. | ||
|
|
||
| Ray Data also provides an Alpha API, `iter_jax_batches`, which streamlines this process by automatically yielding globally sharded JAX Arrays. This can be more efficient and requires less boilerplate code. |
Contributor
There was a problem hiding this comment.
I think we can add a iter_jax_batches api doc in https://docs.ray.io/en/latest/data/api/dataset.html
and then add the link here,
Contributor
Author
There was a problem hiding this comment.
done in d499698, i also modified the API in dataset.py to be part of CD_API_GROUP, so this requires Ray Data review now
liulehui
reviewed
May 12, 2026
liulehui
reviewed
May 12, 2026
…ro_to_jax_trainer guides doc: natively use iter_jax_batches inside intro_to_jax_trainer guides Signed-off-by: Ryan O'Leary <ryanaoleary@google.com>
ryanaoleary
force-pushed
the
doc-iter-jax-batches
branch
from
d499698 to
2a2a976
Compare
![cursor[bot]](https://avatars.githubusercontent.com/in/1210556?s=60&v=4){kind=small}
There was a problem hiding this comment.
Cursor Bugbot has reviewed your changes and found 1 potential issue.
Reviewed by Cursor Bugbot for commit 2a2a976. Configure here.
Signed-off-by: Ryan O'Leary <ryanaoleary@google.com>
ryanaoleary
force-pushed
the
doc-iter-jax-batches
branch
from
1163f7a to
de6cb29
Compare
ray-gardener
Bot
added
docs
liulehui
approved these changes
May 12, 2026
Contributor
Author
|
@bveeramani wondering if you can review from the Ray Data side for the small |
dancingactor
pushed a commit
to dancingactor/ray
that referenced
this pull request
May 13, 2026
…ject#63294) ## Description ray-project#61630 added support for JAX to Ray data, specifically implementing a `iter_jax_batches` util to yield natively sharded `jax.Arrays`. This provides first-class support for processing data within JaxTrainer workloads. This PR updates an existing GPT-2 guide using the `JaxTrainer` to showcase how this new util could simplify the Train code. ## Related issues ray-project#55162 ## Additional information > Optional: Add implementation details, API changes, usage examples, screenshots, etc. --------- Signed-off-by: Ryan O'Leary <ryanaoleary@google.com> Signed-off-by: Ryan O'Leary <113500783+ryanaoleary@users.noreply.github.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
am-kinetica
pushed a commit
to kineticadb/ray
that referenced
this pull request
May 14, 2026
…ject#63294) ## Description ray-project#61630 added support for JAX to Ray data, specifically implementing a `iter_jax_batches` util to yield natively sharded `jax.Arrays`. This provides first-class support for processing data within JaxTrainer workloads. This PR updates an existing GPT-2 guide using the `JaxTrainer` to showcase how this new util could simplify the Train code. ## Related issues ray-project#55162 ## Additional information > Optional: Add implementation details, API changes, usage examples, screenshots, etc. --------- Signed-off-by: Ryan O'Leary <ryanaoleary@google.com> Signed-off-by: Ryan O'Leary <113500783+ryanaoleary@users.noreply.github.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com> Signed-off-by: anindyam1969 <amukherjee@kinetica.com>
Lucas61000
pushed a commit
to Lucas61000/ray
that referenced
this pull request
May 15, 2026
…ject#63294) ## Description ray-project#61630 added support for JAX to Ray data, specifically implementing a `iter_jax_batches` util to yield natively sharded `jax.Arrays`. This provides first-class support for processing data within JaxTrainer workloads. This PR updates an existing GPT-2 guide using the `JaxTrainer` to showcase how this new util could simplify the Train code. ## Related issues ray-project#55162 ## Additional information > Optional: Add implementation details, API changes, usage examples, screenshots, etc. --------- Signed-off-by: Ryan O'Leary <ryanaoleary@google.com> Signed-off-by: Ryan O'Leary <113500783+ryanaoleary@users.noreply.github.com> Co-authored-by: gemini-code-assist[bot] <176961590+gemini-code-assist[bot]@users.noreply.github.com>
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Description
#61630 added support for JAX to Ray data, specifically implementing a
iter_jax_batchesutil to yield natively shardedjax.Arrays. This provides first-class support for processing data within JaxTrainer workloads. This PR updates an existing GPT-2 guide using theJaxTrainerto showcase how this new util could simplify the Train code.Related issues
#55162
Additional information