Cognitive Tasks#
A modular, composable framework for constructing cognitive tasks for neural-network training and computational neuroscience simulations.
What you’ll find here#
A quickstart showing how to sample single trials and JIT/
vmap-friendly batches from a pre-built task.A building-blocks tutorial covering features, declarative phases, encoders, label helpers, and both the instance-based and class-based
Taskdefinition patterns — plus vector-output tasks, conditional control flow, and custom encoders.A variable-length trial sequences tutorial covering
VariableDuration, conditional control flow (If/Switch/While), the packed-buffer-plus-mask layout returned bysample_trialandbatch_sample(..., return_mask=True), masked losses, and theTruncExp/UniformDurationduration samplers.
For the full list of pre-built tasks, encoders, phases, and utilities, see the API reference.