Replace NNI with auto-sklearn, and add some testing #4
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
This probably means that the code will only work run from the blindml source directory. but it's better than not running.
… least tooling as installed by this is broken right now for me in a clean venv - pip-compile is broken because pip-tools breaks with recent pip, and apparently the requirements.txt doesnt pin that stuff hard enough. - the breakage only happens when I try to add in auto-sktools to requirements.txt and rebuild... this is an upgrade of pip-tools and a rerun of the regenerate/reinstall until a fixpoint is reached in my env first manually edit requirements.txt because pip-compile cant do this itself to constrain pip-tools>=5.5 then: run this over and over until no more git diff: (looks like one iteration is ok for me actually) rm -rfv ~/cqx/c/datastation/virtualenv/ && virtualenv /home/benc/cqx/c/datastation/virtualenv/ --python=python3 && pip install -r requirements.txt --extra-index-url https://pypi.anaconda.org/scipy-wheels-nightly/simple && pip-compile --extra-index-url https://pypi.anaconda.org/scipy-wheels-nightly/simple
…roken This gives basic linting for style errors and some basic static semantic checking. Future commits could tidy up the broken flake8 tests and re-enable them. Run with: $ flake8
Remove existing test that fails for me Adjust path of other existing test to run from blindml repo root for when it is invoked by pytest there
makslevental
suggested changes
Feb 26, 2021
blindml/frontend/config/task/task.py
Outdated
Comment on lines
80
to
81
| # ^ should this drop the y_col even when there isn't a drop_cols | ||
| # configuration? |
Contributor
There was a problem hiding this comment.
yes because otherwise we're training on the y_col to predict y_col (i.e. poisoning the training)
Contributor
Author
There was a problem hiding this comment.
i guess if the user specifies the ycol in their explicitly specified X_cols, that's their own fault, though?
Contributor
There was a problem hiding this comment.
yup major party foul (but there should also probably be a check)
Contributor
Author
There was a problem hiding this comment.
it looks like the explicit X_cols code path was broken anyway - I've put in a fix and added a test case.
I've also added in an assert in the explicit X_cols path for "no ycol in xcols"
Contributor
|
Should've put this in the review whoops: please remove references to |
Prior to this, auto-sklearn was only called in regression mode can i get a classification test/demo? does the previous NNI implementation in blindml support classification, to test my changes against? I couldn't get it to work on a simple csv that i made up
…h because of that
benclifford
force-pushed
the
benc-replace-nni-autosklearn
branch
from
92bada1 to
d6ae253
Compare
makslevental
approved these changes
Apr 9, 2021
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: