What can you do with 1000 neurons? Task 2

Dear all, The first task (simple decision) has been solved by a number of participants. Most of the models were able to survive in the proposed environment for quite a long time, and one entry used the optimal decision strategy, with only 22 neurons (out of 1000 available). The challenge is still running with a new task (cued decision, which might be simpler) and a new agenda (you have two months). You'll find instructions at https://github.com/rougier/braincraft the rules for the new task. Reminder: you have have 1000 neurons (leaky rate models), 100 seconds of training time and 10 runs for evaluation. The difficulty comes from the imposed constraints. You have to 2 months to propose your model. Best, Nicolas Rougier. -- Nicolas P. Rougier —— www.labri.fr/perso/nrougier Institute of Neurodegenerative Diseases, Bordeaux

On Wed 03 Sep 2025 at 06:36, "Nicolas P. Rougier (inria) via Comp-neuro" <comp-neuro@lists.cnsorg.org> wrote: Dear all, The challenge is still running with a cued decision task. Provided you've made a model of BG, it might be simple to adapt it, taking into account all the constrains. You'll find instructions at hdttps://github.com/rougier/braincraft Reminder: you have have 1000 neurons (leaky rate model), 100 seconds of training time and 10 runs for evaluation. The difficulty comes from the imposed constrains. You now have to 2 weeks to propose your model. Best, Nicolas Rougier. -- Nicolas P. Rougier —— www.labri.fr/perso/nrougier Institute of Neurodegenerative Diseases, Bordeaux
participants (1)
-
Nicolas P. Rougier (inria)