v0.1.3

LaplaceRedux v0.1.3

Diff since v0.1.2

Closed issues:

• Add support for block-diagonal Hessian approximations (#12)
• Implement Generalized Gauss Newton (GGN) approximation (#17)
• Interface MLJ (#18)
• Add support for mini-batch training (#19)
• Refactor gradient and jacobians as multi-dimensional arrays (#20)
• Update Documentation (#22)
• 🏃🏽 Getting started (#24)
• 🚀 Lift off (#25)
• 🌯 Wrapping Up (#26)
• Add support for Last-Layer and Subnet Laplace (#27)
• Organise code base (#36)
• Move FormatCheck into separate file (#40)

Merged pull requests:

• 18 interface mlj (#21) (@pat-alt)
• Clean README, docs and docstrings (#23) (@pitmonticone)
• Complete CSE2000 contributions (#33) (@severinbratus)
• Improve docstrings (#35) (@severinbratus)
• FormatCheck into separate file (#41) (@pat-alt)
• sorting out compat issues (#56) (@pat-alt)
• CompatHelper: bump compat for Flux to 0.14, (keep existing compat) (#57) (@github-actions[bot])
• CompatHelper: bump compat for MLJFlux to 0.3, (keep existing compat) (#58) (@github-actions[bot])