This repository has been archived by the owner on Jul 7, 2023. It is now read-only.
Create Text2RealProblem class for regression-based problems #1748
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Outdated suggestions cannot be applied.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.
A variety of NLP problems involve training regression models to predict continuous-valued ratings from text (e.g., sentiment scoring). While T2T supports several
REAL
modalities, there is no base class for regression-based text problems. This PR introduces aText2RealProblem
that is analogous to the existingText2ClassProblem
, but for regression problems.The class also supports multitask regression via the
ntasks
property, which allows for the regression target to be an n-dimensional float vector.Note that this PR doesn't contribute any additional problems to the registry (for now), since the datasets I've been training on are not public. That said, this base class makes it very straightforward to define a custom text regression problem, and I hope that it will be useful for other T2T users.
@lukaszkaiser This PR is very similar to the Gist you reviewed from me -- please let me know what you think!