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

Performance improvements and fixes #6

Merged
merged 31 commits into from
Jan 8, 2024
Merged

Performance improvements and fixes #6

merged 31 commits into from
Jan 8, 2024

Conversation

davidnabergoj
Copy link
Owner

This PR fixes/improves speed of operations for invertible residual networks, residual flows, proximal residual flows, OTFlow, FFJORD, RNODE. It also includes a basic stochastic variational inference fitting procedure.

@davidnabergoj
Fix device
6eec626
@davidnabergoj
Fix weight shape
76db053
@davidnabergoj
Change defaults for LRS
304092e
@davidnabergoj
Change default bin size in LRS
f0167a3
@davidnabergoj
Reduce number of tests for faster CI/CD
9f22ac8
@davidnabergoj
Rewrite elementwise bijection to avoid null conditioner
8ac96f8
@davidnabergoj
Remove Conditioner and NullConditioner classes
8a8572b
@davidnabergoj
Rename files
cc607f6
@davidnabergoj
Towards fixing autograd in residual flows
c0e239e
@davidnabergoj
Fix autograd for InvertibleResNet
6a17859
@davidnabergoj
Fixing autograd for ResFlow
952b3a2
@davidnabergoj
Create graph in roulette estimator only if training
3df7362
@davidnabergoj
Set default hidden layer size of spectral neural network to max(log(n…
43168a5 
…_dim), 4)
@davidnabergoj
Change default hidden layer size in proximal neural network
a50432a
@davidnabergoj
Set default hidden size in OTFlow
8ee1e7c
@davidnabergoj
Fix OTFlow forward output sign
e82e79a
@davidnabergoj
Handle training=True for residual bijections in Flow.fit
9624725
@davidnabergoj
Fix divergence sign for FFJORD, RNODE
549e5c0
@davidnabergoj
Fix gamma in Proximal ResFlow with single layer blocks
c8be5c1
@davidnabergoj
Change defaults for invertible ResNet and ResFlow
b736d6f
@davidnabergoj
Fix log determinant in residual bijection, interleave layers with ele…
abb32dc 
…mentwise affine maps
@davidnabergoj
Use different residual blocks in Invertible ResNet and ResFlow
2560814
@davidnabergoj
Change ProximalResFlow defaults
5aaf37a
@davidnabergoj
Fix Gaussian quadrature call for UMNN-MAF
fd4dad8
@davidnabergoj
Better skip message for residual bijection reconstruction test
3e708b1
@davidnabergoj
Change UMNN MAF parametrization
3026654
@davidnabergoj
Add variational fit
8dcea8b
@davidnabergoj
Fix SVI, add docs
b17da39
@davidnabergoj
Skip UMNN tests
9b87f8f
@davidnabergoj
Merge branch 'dev' of https://github.com/davidnabergoj/normalizing-flows
19ddb80 
 into dev

# Conflicts:
#	test/test_umnn.py
@davidnabergoj davidnabergoj merged commit 7b83bd0 into master Jan 8, 2024
5 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers
No reviews
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
1 participant
@davidnabergoj