Move to datasets & apply interface #685
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
This PR is the first one of multiple PRs to replace #665. This PR only focuses on implementing the new pipeline interface, without adding any new functionality. The new interface applies operations to intermediate datasets instead of adding operations to a pipeline, as shown below. It's a superficial change, since only the interface is changed. All underlying behavior is still the same. The new interface fits nicely with our data format design and we'll be able to leverage it for interactive development in the future. We can calculate the schema for each intermediate dataset so the user can inspect it. Or with eager execution, we could execute a single operation and allow the user to explore the data using the dataset. I still need to update the README generation, but I'll do that as a separate PR. It becomes a bit more complex since we now need to discriminate between read, transform, and write components to generate the example code. **Old interface** ```Python from fondant.pipeline import ComponentOp, Pipeline pipeline = Pipeline( pipeline_name="my_pipeline", pipeline_description="description of my pipeline", base_path="/foo/bar", ) load_op = ComponentOp( component_dir="load_data", arguments={...}, ) caption_op = ComponentOp.from_registry( name="caption_images", arguments={...}, ) embed_op = ComponentOp( component_dir="embed_text", arguments={...}, ) write_op = ComponentOp.from_registry( name="write_to_hf_hub", arguments={...}, ) pipeline.add_op(load_op) pipeline.add_op(caption_op, dependencies=[load_op]) pipeline.add_op(embed_op, dependencies=[caption_op]) pipeline.add_op(write_op, dependencies=[embed_op]) ``` **New interface** ```Python pipeline = Pipeline( pipeline_name="my_pipeline", pipeline_description="description of my pipeline", base_path="/foo/bar", ) dataset = pipeline.read( "load_data", arguments={...}, ) dataset = dataset.apply( "caption_images", arguments={...}, ) dataset = dataset.apply( "embed_text", arguments={...}, ) dataset.write( "write_to_hf_hub", arguments={...}, )
This PR updates the component generation to take into account the component type and generate the appropriate usage example.
PR that introduces functionality to new pipeline interface as discussed [here](#567 (comment)) * The component spec now accepts **OneOf** additionalFields or Fields in it's consumes and produces section * The new `consumes` and `produces` are defined at the Op level similarly to the ones in the component spec, if they are present, they will override the default `consumes` and `produces` defined in the component spec (manifet, dataIO) * Some changes were added to `DataIO` just to resolve tests issues but the new functionality of the custom consumes and produces is not implemented yet (will be tackled in a separate PR) --------- Co-authored-by: Robbe Sneyders <robbe.sneyders@ml6.eu>
Follows #695 This PR adds the mapping functionality the `dataIO` module --------- Co-authored-by: Robbe Sneyders <robbe.sneyders@ml6.eu>
RobbeSneyders
force-pushed
the
feature/dataset-apply-interface
branch
from
200d06a
to
f5174a1
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.
This is a feature branch to collect all changes related to the datasets & apply interface until we have a state we can merge.