Releases: impICNF/ContinuousNormalizingFlows.jl
Releases · impICNF/ContinuousNormalizingFlows.jl
v0.29.3
ContinuousNormalizingFlows v0.29.3
Merged pull requests:
- Bump peter-evans/create-pull-request from 7 to 8 (#498) (@dependabot[bot])
- Bump julia-actions/julia-format from 3 to 4 (#499) (@dependabot[bot])
- Bump actions/checkout from 4 to 6 (#500) (@dependabot[bot])
v0.23.1
v0.23.0
v0.29.2
ContinuousNormalizingFlows v0.29.2
v0.29.1
ContinuousNormalizingFlows v0.29.1
Merged pull requests:
- CompatHelper: bump compat for Optimization to 5, (keep existing compat) (#494) (@github-actions[bot])
v0.29.0
ContinuousNormalizingFlows v0.29.0
Breaking changes
-
now
ICNFModelgetsepochsassol_kwargs(to be able to use any optimizer) -
now
ICNFModelgets the batch size asbatchsizeinstead ofbatch_size(to be similar to other packages naming)
Merged pull requests:
v0.28.2
ContinuousNormalizingFlows v0.28.2
v0.28.1
v0.28.0
ContinuousNormalizingFlows v0.28.0
Breaking changes
-
use_batch = falsereplaced withbatch_size = 0in MLJ models. -
Augmented variables get regularized by default now. (only when n.augmented >= n.actual)
-
Return solve stats instead of printing them in MLJ models.
Merged pull requests:
- Test with
ForwardDiff(#466) (@prbzrg) - CompatHelper: bump compat for ForwardDiff to 1 for package test, (keep existing compat) (#467) (@github-actions[bot])
- test default kwargs (#468) (@prbzrg)
- CompatHelper: bump compat for JET to 0.10 for package test, (keep existing compat) (#469) (@github-actions[bot])
- Simplify Tests (#472) (@prbzrg)
- CompatHelper: bump compat for DifferentiationInterface to 0.7, (keep existing compat) (#473) (@github-actions[bot])
- CompatHelper: bump compat for DifferentiationInterface to 0.7 for package test, (keep existing compat) (#474) (@github-actions[bot])
- CompatHelper: bump compat for DifferentiationInterface to 0.7 for package benchmark, (keep existing compat) (#475) (@github-actions[bot])
v0.27.0
ContinuousNormalizingFlows v0.27.0
Breaking changes
- To use GPUs, we use
MLDataDevicesnow instead of handling it viaComputationalResources.
Merged pull requests: