Use vectorized jacobian in minimize ops #1558
Draft
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
ricardoV94
force-pushed
the
use_vectorized_jacobian_in_minimize_ops
branch
from
53721c4 to
f1a8ae1
Compare
|
This is hitting #1550 I don't know if because I introduced a bug that triggers it or what not. Let's fix that first and then come back to this |
Also cover cases of AdvancedIncSubtensor with batch indices that were not supported before
ricardoV94
force-pushed
the
use_vectorized_jacobian_in_minimize_ops
branch
from
f1a8ae1 to
7a7ad7a
Compare
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR uses the more new experimental vectorized jacobian in the minimize/root Ops, with the goal of getting it more tested, as it won't be breaking any people's code.
The existing notebook revealed we were missing some rewrites to avoid Blockwise(IncSubtensor), that this PR also addresses. In the printed example we can see that the vectorized jacobian replaces a Scan with two IncSubtensors by two IncSubtensors on an expanded buffer. This should be much more performant for this sort of small graphs.
Either my chages introduce or reveal a bug on the notebook, so this is still a draft
📚 Documentation preview 📚: https://pytensor--1558.org.readthedocs.build/en/1558/