<!-- Release notes generated using configuration in .github/release.yml at main --> What's Changed New Features 🎉 Add full mlx blockwise support by @jessegrabowski in https://github.com/pymc-devs/pytensor/pull/2233 Bugfixes 🐛 Fix numba CauchyRV loc/scale by @notluquis in https://github.com/pymc-devs/pytensor/pull/2309 Maintenance 🔧 Add rewrite for diag(A @ B) to avoid materializing the product by @jessegrabowski in https://github.com/pymc-devs/pytensor/pull/2302 Implement erfcx and erfcinv in JAX without tensorflow-probability by @andreshernandez-spec in https://github.com/pymc-devs/pytensor/pull/2318 Stable logsumexp softmax gradients without rewrites by @ricardoV94 in https://github.com/pymc-devs/pytensor/pull/2313 Bump numba upper bound to <=0.66.0 (main) by @github-actions[bot] in https://github.com/pymc-devs/pytensor/pull/2273 New Contributors @notluquis made their first contribution in https://github.com/pymc-devs/pytensor/pull/2309 @andreshernandez-spec made their first contribution in https://github.com/pymc-devs/pytensor/pull/2318 Full Changelog: https://github.com/pymc-devs/pytensor/compare/rel-3.2.3...rel-3.2.4