Change features vs schema logic #423
Merged
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Outdated suggestions cannot be applied.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.
New logic for
nlp.Featuresin datasetsPreviously, it was confusing to have
featuresand pyarrow'sschemainnlp.Dataset.However
featuresis supposed to be the front-facing object to define the different fields of a dataset, whileschemais only used to write arrow files.Changes:
schemafield innlp.Datasetfeaturesthe source of truth to read/write examplesfeaturescan no longer beNoneinnlp.Datasetfeaturesafter each dataset transform such asnlp.Dataset.mapTodo: change the tests to take these changes into account