Updated Notebook to use new dataclasses implementation for setting SSM information#859
Conversation
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
|
View / edit / reply to this conversation on ReviewNB jessegrabowski commented on 2026-03-16T02:55:19Z Line #10. def set_states(self): half as long! Dekermanjian commented on 2026-03-16T21:27:00Z Yeah, it really is so much better this way! |
|
View / edit / reply to this conversation on ReviewNB jessegrabowski commented on 2026-03-16T02:55:20Z 60s -> 20s on sampling is nice too (unless you got a new computer) Dekermanjian commented on 2026-03-16T21:31:47Z I am using the same computer. The speed-boost is really nice. Is this due to the inplace blockwise work you did in PyTensor?
|
|
View / edit / reply to this conversation on ReviewNB jessegrabowski commented on 2026-03-16T02:55:21Z The Dekermanjian commented on 2026-03-16T21:32:14Z I will take a look at how the new API does this and fix it. |
There was a problem hiding this comment.
Looks great! Someone else will have to weigh in on why we have merge conflicts. Maybe @fonnesbeck knows?
|
Yeah, it really is so much better this way! View entire conversation on ReviewNB |
|
I am using the same computer. The speed-boost is really nice. Is this due to the inplace blockwise work you did in PyTensor?
View entire conversation on ReviewNB |
|
I will take a look at how the new API does this and fix it. View entire conversation on ReviewNB |
Dekermanjian
force-pushed
the
ssm_hurricane
branch
from
32e96b6 to
831b94f
Compare
2 participants
When setting an SSM's states, parameters, exogenous data, shocks, and coordinates we now have a new implementation using
dataclasses. I updated the models in the notebook to use this new design.