Add GenerateSentencePair task
#689
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
gabrielmbmb
force-pushed
the
embedding-dataset-tasks
branch
from
05b3886
to
fd9d64a
Compare
CodSpeed Performance ReportMerging #689 will not alter performanceComparing Summary
|
gabrielmbmb
force-pushed
the
embedding-dataset-tasks
branch
from
fd9d64a
to
2e5d9f8
Compare
alvarobartt
reviewed
Jun 3, 2024
gabrielmbmb
force-pushed
the
embedding-dataset-tasks
branch
from
d1e00be
to
06924a9
Compare
alvarobartt
reviewed
Jun 4, 2024
alvarobartt
reviewed
Jun 4, 2024
| @@ -30,17 +30,15 @@ | |||
| from distilabel.steps.tasks.prometheus_eval import PrometheusEval | |||
| from distilabel.steps.tasks.quality_scorer import QualityScorer | |||
| from distilabel.steps.tasks.self_instruct import SelfInstruct | |||
| from distilabel.steps.tasks.sentence_transformers import GenerateSentencePair | |||
| from distilabel.steps.tasks.structured_generation import StructuredGeneration | |||
| from distilabel.steps.tasks.text_generation import ChatGeneration, TextGeneration | |||
| from distilabel.steps.tasks.typing import ChatItem, ChatType | |||
| from distilabel.steps.tasks.ultrafeedback import UltraFeedback | |||
|
|
|||
| __all__ = [ | |||
There was a problem hiding this comment.
The import order here was alphabetical, what's the rationale behind this change? Maybe we should change this in other places too to make sure we're aligned?
There was a problem hiding this comment.
The rationale was to have the imports ordered in __all__ by the order in which the were imported, which is more common than having them alphabetically.
alvarobartt
reviewed
Jun 4, 2024
alvarobartt
reviewed
Jun 4, 2024
alvarobartt
reviewed
Jun 4, 2024
Co-authored-by: alvarobartt <[email protected]>
gabrielmbmb
force-pushed
the
embedding-dataset-tasks
branch
from
03aa740
to
3a1be83
Compare
Co-authored-by: alvarobartt <[email protected]>
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.
Description
This PR adds a new task called
GenerateSentencePairthat allows building datasets that can be used to train embedding models. The task can be used to generate apositivesentence based on the providedanchorsentence, and iftripletattribute isTrue, then it will generate anegativesentence too. The task can be used to paraphrase the anchor, generate semantically similar content with respect to the anchor, or to generate a query for the anchor.In addition, this PR has updated the
add_raw_outputattribute so it's now aRuntimeParameter, and it has nowTrueas default value, so the raw outputs of the LLMs are stored by default in the final dataset.