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.

Sign up for GitHub

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

Jump to bottom

✨ Enable regression & misc #27

Merged
merged 73 commits into from
Jul 8, 2023
Merged

✨ Enable regression & misc #27

merged 73 commits into from
Jul 8, 2023

Conversation

o-laurent
Copy link
Contributor

@o-laurent o-laurent commented Jun 13, 2023 •
edited
Loading

  • Compress GNNL and NNL
  • Make the MLP more flexible
  • Do we really need to add a parameter to the CLI (regression vs. classification)? Why not use dm for this?
  • Add VGGs
  • Add LeNet
  • Add a setup.py when poetry is not available + mention it on the installation page
  • Improve the MLP baseline
  • Add ensemble regression
  • How to deal with mutlivariate regression cases? (different output) -> Ask for the dimension first
  • Fix Bayesian layers? (cf failed tests: 🐛 Bayesian layers may fail during tests #28)
  • Add bayesian models & losses
  • Add deep ensembles method
  • Add a Bayesian tutorial
  • Fix tutorial generation
  • Make tutorial gallery static
  • Add a pretrained tutorial
  • Add a regression tutorial
  • Add tests
  • Add docs
  • Improve VGG optim recipes?

Done for now:
Add an arg to CLI to split regression and cls 🔨 Modify tests accordingly ✔️
Add a regression recipe ✨
Add a small MLP model ✨
Add an MLP baseline ✨
Add a UCIRegression Datamodule ✨
Add GaussianNLL lightning metric ✨
Add an experiment ✨

@o-laurent o-laurent self-assigned this Jun 13, 2023
@o-laurent
Start the regression branch ✨
e4c00ec 
Add an arg to CLI to split regression and cls 🔨
Modify tests accordingly :check_mark:
Add a regression recipe ✨
Add a small MLP model ✨
Add an MLP baseline ✨
Add a UCIRegression Datamodule ✨
Add GaussianNLL lightning metric ✨
Add an experiment ✨
@o-laurent
Remove GNNL duplicate code 🔥
44e552a
@o-laurent
Fix validation split arg. 🐛
2fca85e
@o-laurent
Fix UCIRegression dataset 🐛
678990e
@o-laurent
Add energy appliances to UCI ✨
ad81902
@o-laurent
Add *.out to .gitignore 🔧
f3c5130
@o-laurent
Improve datamodules' docstrings 📚
df77a53
@o-laurent
Improve UCI datamodule ⚡
1695530
@o-laurent
Enable standard regression ✨
52ae849
@o-laurent o-laurent changed the title Start the regression branch ✨ Enable regression training ✨ Jun 16, 2023
@o-laurent o-laurent changed the title Enable regression training ✨ Enable regression training & misc ✨ Jun 23, 2023
@o-laurent o-laurent changed the title Enable regression training & misc ✨ Enable regression & misc ✨ Jun 23, 2023
@o-laurent o-laurent marked this pull request as ready for review July 4, 2023 14:49
@o-laurent
Copy link
Contributor Author

o-laurent commented Jul 4, 2023 •
edited
Loading

Documentation is missing but the code seems approx fine.
I'm working on the doc now.

@o-laurent o-laurent requested a review from alafage July 4, 2023 14:50
@o-laurent o-laurent changed the title Enable regression & misc ✨ ✨ Enable regression & misc Jul 4, 2023
o-laurent and others added 13 commits July 4, 2023 20:18
@o-laurent
🎨 Continue improve typing & misc
a2d44b8
@o-laurent
🐛 Fix gallery intro
c8d30f0
@o-laurent
🔥 Remove empty test file
2fff0ed
@o-laurent
✨ cli_main returns test results
62b5f12 
🎨 rename cls_main to cli_main
@o-laurent
✨ Add CIFAR-N
f9f955e
@o-laurent
🐛 Fix non-dist ensemble regression
e0d1d2f
@o-laurent
🎨 Misc
48a7f94
@o-laurent
🐛 Fix bayesian conv
b3abd8d
@alafage
🐛 Fix BrierScore update when handling binary classification
9dd884f
@alafage
🐛 Fix DeepEnsembles model
287e3f8 
models are now wrapped into a ModuleList enabling efficient device migration and parameter registration
@alafage
✨ DeepEnsemble baseline on classification tasks
adf40ba 
tests have been made on CIFAR10 with two vanilla ResNet18 using the associated experiment file
@alafage
➕ pandas is now a dependency of torch-uncertainty
359d682
@alafage
✨ DeepEnsemble baseline on regression tasks
dd357c7 
tests done using two MLPs on kin8nm dataset.
o-laurent
o-laurent commented Jul 5, 2023
View reviewed changes
experiments/classification/mnist/bayesian_lenet.py Outdated Show resolved Hide resolved
torch_uncertainty/optimization_procedures.py Show resolved Hide resolved
torch_uncertainty/models/vgg/packed.py Show resolved Hide resolved
torch_uncertainty/models/utils.py Outdated Show resolved Hide resolved
torch_uncertainty/models/lenet.py Outdated Show resolved Hide resolved
torch_uncertainty/metrics/nll.py Show resolved Hide resolved
torch_uncertainty/layers/bayesian_layers/sampler.py Show resolved Hide resolved
torch_uncertainty/layers/bayesian_layers/bayes_linear.py Outdated Show resolved Hide resolved
torch_uncertainty/datasets/uci_regression.py Show resolved Hide resolved
torch_uncertainty/baselines/regression/mlp.py Outdated Show resolved Hide resolved
alafage
alafage requested changes Jul 8, 2023
View reviewed changes
Copy link
Contributor

@alafage alafage left a comment

Choose a reason for hiding this comment

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

It seems to me that there are typos in some datamodules when defining both train and validation sets using random_split. Otherwise this pull request is fine enough to be merged to dev :)

torch_uncertainty/datamodules/cifar10.py Outdated Show resolved Hide resolved
@alafage alafage merged commit 7d4aecb into dev Jul 8, 2023
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@alafage alafage alafage requested changes

Assignees

@o-laurent o-laurent

Labels
enhancement New feature or request
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@o-laurent @alafage