This repository has been archived by the owner on Dec 16, 2022. It is now read-only.
added AdversarialBiasMitigator tests and model #281
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
|
I'm not reviewing this until allenai/allennlp#5269 is merged, right? |
dirkgr
reviewed
Jun 17, 2021
| "display_name": "Adversarial Binary Gender Bias-Mitigated RoBERTa SNLI", | ||
| "task_id": "textual_entailment", | ||
| "model_details": { | ||
| "description": "This `Model` implements a basic text classifier and feedforward regression adversary with an adversarial bias mitigator wrapper. The text is embedded into a text field using a RoBERTa-large model. The resulting sequence is pooled using a cls_pooler `Seq2VecEncoder` and then passed to a linear classification layer, which projects into the label space. Subsequently, a `FeedForwardRegressionAdversary` attempts to recover the coefficient of the static text embedding in the binary gender bias subspace. While the adversary's gradients are computed normally, the classifier's gradients are computed such that updates to the classifier's parameters will not aid the adversary and will make it more difficult for the adversary to recover binary gender bias.", |
There was a problem hiding this comment.
If this method really doesn't work that well, as you said in standup, then I think we should have a description of that somewhere.
There was a problem hiding this comment.
The method actually ended up working quite well.
added 2 commits
June 17, 2021 22:12
…om/allenai/allennlp-models into arjuns/adversarial-bias-mitigation
ArjunSubramonian
changed the title
AkshitaB
approved these changes
Jun 18, 2021
| "preprocessing": null | ||
| }, | ||
| "quantitative_analyses": { | ||
| "unitary_results": null, |
There was a problem hiding this comment.
I'm curious; what numbers did we get?
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
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.
AdversarialBiasMitigatortests.adversarial-binary-gender-bias-mitigated-roberta-snlimodel.Depends on: allenai/allennlp#5269.