Blog

• August 2026 JAX kernel fusion — Evaluating JAX kernel fusion using XLA, Pallas and CUDA FFI.
• July 2026 Blaxbird v0.2.0 — Sharding NNX models over an arbitrary mesh.
• June 2026 Learning MLX — Learning Apple's array framework MLX by implementing a compute-heavy ABC sampler.
• June 2026 🦉🎉🦉Relaunching my blog🦉🎉🦉
• January 2023 Diffusion models II — Introducing a new class of generative models using stochastic differential equations and denoising score matching.
• November 2022 Probabilistic reconciliation — On two recent methods on reconciliation of hierarchical time series forecasts.
• June 2022 Diffusion models I — Introducing a novel class of generative models that are inspired by non-equilibrium thermodynamics.
• May 2022 Reverse-mode autodiff in Clojure — A proof-of-concept autodiff engine built to learn Clojure.
• January 2022 Normalizing flows for variational inference — Implementing an *inverse autoregressive flow* for variational inference of parameters in a simple bivariate Gaussian example in Jax, Distrax, Optax and Haiku.
• September 2021 Tensor-product spline smoothers — Implementing a probabilistic model for causal inference with structured latent confounders.
• August 2021 VI for stick-breaking constructions — Implementing mean-field variational approximations for nonparametric mixture and factor models using stick-breaking constructions.
• July 2021 Hilbert-space approximate copula processes — Explaining how a copula process in conjunction with Hilbert-space approximations can be used to model stochastic volatility.
• June 2021 Variational LSTMs — Implementing a variational multivariate LSTM for timeseries prediction of an US election data set.
• May 2021 Hierarchical, coregionalized GPs — Implementing two GP models and compares their predictive performance as well as MCMC diagnostics on an US election data set.
• March 2021 Bayesian optimization — The basics of optimization of costly functions with probabilistic surrogate models.
• October 2020 Normalizing flows — How TensorFlow Probability can be used to implement a custom normalizing flow.
• June 2020 Causal structure learning using VAEs — Implementing a novel graph variational autoencoder and compares it to *greedy equivalence search*, one of the state-of-the-art methods for causal discovery.
• January 2020 Sequential regression models — Introducing a special class of ordinal regression models which assume a sequential response mechanism.
• October 2019 Structure MCMC — How PyMC3 can be used to learn the structure of a Bayesian network.
• October 2019 Mixed models — shows concise reference implementations for optimization of the objective of (generalized) linear mixed models.
• June 2019 SBC — On validating Bayesian posterior inferences.
• January 2019 Dirichlet process mixture models — Extending the Bayesian mixture to the infinite setting.
• October 2018 Gaussian process classification — Extending Gaussian process regression to classification scenarios.
• October 2018 Gaussian process regression — Non-parametric Bayesian regression.