search

lnccbrown / HSSM

Development of HSSM package
Other
71 stars 10 forks source link

Implementing RLDDM in HSSM #386

Open theonlydvr opened 3 months ago

theonlydvr commented 3 months ago

I previously was heavily using the HDDM package to fit RLDDMs and I was hoping to continue this work with the added flexible functionality of HSSM. However, from what I can see, these models have not been added to the package and the issue on them was closed with no comment.

I was wondering what the status of these models in the package was? If there's no internal effort to port these models, I would be happy to work on the implementation if someone could let me know the features that would be necessary for such a model in the HSSM framework.

frankmj commented 3 months ago

Great question as we have been making a decent amount of progress on that front - to not only include RLDDM but (i) to make it flexible so that the user can specify any RL model (e.g. via a blackbox likelihood) and (ii) to make the RL models differentiable so we can leverage the NUTS sampler. Progress has been made on these fronts and we should be ready to include them in one of the next releases and would certainly welcome some test users (which could include you adding your own models etc). cc'ing others in this thread that are working on this in my group

M

Michael J Frank, PhD | Edgar L. Marston Professor Director, Carney Center for Computational Brain Science https://www.brown.edu/carney/ccbs Laboratory of Neural Computation and Cognition https://www.lnccbrown.com/

Brown University website http://ski.clps.brown.edu

On Tue, Apr 2, 2024 at 5:05 PM theonlydvr @.***> wrote:

I previously was heavily using the HDDM package to fit RLDDMs and I was hoping to continue this work with the added flexible functionality of HSSM. However, from what I can see, these models have not been added to the package and the issue on them was closed with no comment.

I was wondering what the status of these models in the package was? If there's no internal effort to port these models, I would be happy to work on the implementation if someone could let me know the features that would be necessary for such a model in the HSSM framework.

— Reply to this email directly, view it on GitHub https://github.com/lnccbrown/HSSM/issues/386, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAG7TFBXQJQOIFJOCZ66WMLY3MMRJAVCNFSM6AAAAABFUDIJJGVHI2DSMVQWIX3LMV43ASLTON2WKOZSGIZDCNBUGIZDONY . You are receiving this because you are subscribed to this thread.Message ID: @.***>

theonlydvr commented 3 months ago

Perfect! I would definitely be interested in testing the new features when they're released!

AlexanderFengler commented 3 months ago

@theonlydvr,

as @frankmj suggested, we are actively working on this. I will mention this during our next team meeting to figure if we can utilize your help at this stage.

Thank you.

Best, Alex