Rémi Louf
Rémi Louf
Should allow us catch issues such as #223.
[Aeppl](https://github.com/aesara-devs/aeppl) is a very expressive PPL written in `aesara`. [Aehmc](https://github.com/aesara-devs/aehmc) implements a NUTS sampler identical to blackjax's and it would be interesting for me to have a benchmark. It would...
From this [paper](https://arxiv.org/pdf/1902.03932.pdf). All the necessary ingredients are here and in Optax. This would give us a deep learning oriented example while demonstrating the flexibility of the kernel design. There...
Metropolis-adjusted Langevin Trajectories. https://arxiv.org/pdf/2202.13230.pdf
We should implement Ensemble MCMC (https://arxiv.org/abs/1801.09065) as it has the potential to speed-up inference on GPU. @AdrienCorenflos shared the following one-file implementation that whoever implements this in `blackjax` can reuse:...
[See the notebook](https://blackjax-devs.github.io/blackjax/examples/HierarchicalBNN.html)
Benchmarks should serve three functions. First prevent performance regression when changes are made on the libray's core. Then give potential users an idea of what to expect when they use...