Conversation
|
@aloctavodia you can see the CI running on my fork juanitorduz#1 The tests |
|
Actually, the changes proposed ensure the test run and they are all green 🟢 🙌 juanitorduz@89581c0 |
distributions/optimization.py
Outdated
| Evaluates bounds at graph-build time for constants. | ||
| """ | ||
| try: | ||
| lower_t = pt.as_tensor_variable(lower) |
There was a problem hiding this comment.
I think we can safely assume everything is a tensor here, and then other libraries than depend on this one could make other assumptions
There was a problem hiding this comment.
So far I have c78a16c . If I do pure symbolic the test keep being stuck ¯\(ツ)/¯
|
coming back to this one soon 🙈 |
|
@aloctavodia I have looked into the feedback and the CI is green now: juanitorduz#1 The hardest part has been able to run the test in a reasonable time so that they do not get stuck. Feel free to push directly on this branch to improve if you see edge cases or feedback if there is a better direction :) |
| [ | ||
| [0.0, 1.0], # Standard logit-normal (centered) | ||
| [0.0, 0.001], # Narrower distribution | ||
| [0.0, 0.5], # Narrower distribution (sigma=0.001 is too extreme for numerical integration) |
There was a problem hiding this comment.
What error do we get? 0.001 is not that extreme; we should be able to handle it easily. If not, we need to improve the integration routine. Asking just to open an issue, not that we need to fix it now.
There was a problem hiding this comment.
Update I may have a fix for this
There was a problem hiding this comment.
great! do you wanna push it in a different PR?
|
In the future, we may want to test compiled and cached functions. |
2 participants
Add CI and some cleaning :)