KPPax compiles KPP chemistry mechanisms into differentiable JAX code
On POLLU, online fine-tuning cuts biogenic emission-correction error by a factor of 2.4 to 5.
6 stories, newest first · one card per story
On POLLU, online fine-tuning cuts biogenic emission-correction error by a factor of 2.4 to 5.
Double-Stitch learns Wasserstein Lagrangian population dynamics simulation-free, training 4-14x faster than solver-based WLM while covering periodic dynamics.

A coding guide demonstrates Google Research's Kauldron, a modular JAX trainer using plain-data configs and string wiring.
Open-source e3j speeds Euclidean-equivariant GPU and TPU kernels, beating cuEquivariance on a MACE simulation.

Sakana AI's PC-ALM adds per-layer Lagrange multipliers to predictive coding, matching backprop on networks up to 1000 layers with layer-local updates.

MarkTechPost tutorial implements a hierarchical NeRF in JAX using jax3d volume-rendering primitives for novel-view synthesis and 3D reconstruction.
Today 1.6K readers · 2.4K views · 775 reads · all time 189K views