Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Feature/499 datainf #582

Merged
merged 44 commits into from
Jun 6, 2024
Merged

Feature/499 datainf #582

merged 44 commits into from
Jun 6, 2024

Conversation

schroedk
Copy link
Collaborator

@schroedk schroedk commented May 21, 2024

Description

This PR closes #499 closes #585

Changes

Checklist

  • Wrote Unit tests (if necessary)
  • Updated Documentation (if necessary)
  • Updated Changelog
  • If notebooks were added/changed, added boilerplate cells are tagged with "tags": ["hide"] or "tags": ["hide-input"]

schroedk added 18 commits May 21, 2024 16:54
* move types to influence.types
* add fit_required decorator
* add class ComposableInfluence, based on generic types from influence.types
* add class SumAggregator to influence.array
…tion_model, implement InverseHarmonicMeanInfluence and add corresponding test
@schroedk schroedk force-pushed the feature/499-datainf branch from b0e297a to 531aacc Compare May 22, 2024 15:40
schroedk added 11 commits May 22, 2024 19:45
* move operator submodules to influence.torch level
* move implementations of generic classes to influence.torch.base
* improve naming
* add and extend documentation
* simplify classes
* renaming classes
* add cocept 'TensorDictOperator', which can act on tensor dictioniaries, to avoid
  intermeditate flatten and concat to reduce memory consumption
@schroedk schroedk marked this pull request as ready for review May 29, 2024 16:11
@schroedk schroedk changed the title WIP: Feature/499 datainf Feature/499 datainf May 29, 2024
@schroedk schroedk added this to the v0.11.0 milestone May 29, 2024
Copy link
Collaborator

@janosg janosg left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think the new abstractions in the types module are really good and will also work well for JAX. The rest are minor comments.

src/pydvl/influence/torch/base.py Show resolved Hide resolved
src/pydvl/influence/torch/base.py Show resolved Hide resolved
src/pydvl/influence/torch/util.py Show resolved Hide resolved
src/pydvl/influence/torch/util.py Outdated Show resolved Hide resolved
tests/influence/torch/test_batch_operation.py Outdated Show resolved Hide resolved
@schroedk
Copy link
Collaborator Author

schroedk commented Jun 5, 2024

Fix issue in GradientProvider: not exposing dict structure of gradients, as this might be framework specific.

Done

@schroedk schroedk requested a review from janosg June 5, 2024 16:42
@schroedk schroedk merged commit 347b55a into develop Jun 6, 2024
35 checks passed
@schroedk schroedk deleted the feature/499-datainf branch June 6, 2024 08:30
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
2 participants