* Refactor modules pruning to filter and latent_estimation to count
* Remove polyplex (research) algorithms from cleanlab
* Create new module rank and move scoring functions to rank.
* Rename test to match new module names
* Fixed error in normalized margin. added ranking for arbitrary psx and labels.
* Remove unused tests and methods. add multi-label support for baseline.
* Move baseline methods to filter and delete baseline module.
* change filter.get_noise_indices to filter.find_label_issues
* Rename baseline methods. fill out docstrings.
* Only require 1 example to be left in each class after removing errors. (instead of 5)
* Remove K as a parameter to count.compute_confident_joint
* Add C_argmax and C_ij methods from CL paper to find_label_issues
* Add warnings for new prune methods and frac_noise. Fix tests.
* Add baseline tests to test_rank_filter and delete baseline test
* Remove inverse_noise_matrix parameter in classification call to find_label_issues
* add todo to update docstring with new ranking functions
* 100% tests pass. add multi-label support for prune_method
* Major NOT-backwards-compatible name changes to most components
* More Major NOT-backwards-compatible name changes
* fixed s -> label mistakes
* Several nomenclature updates from PR feedback. models renamed to example models.
* Remove python2 support across all modules.
* major api changes. psx -> pred_probs. prob_given_label -> self_confidence. testing added.
* enable python version 3.9 for pytorch model.
* ran spellcheck
* ran grammar check
* Update count.py
* Update filter.py
* Update setup.py and ci.yml to no longer support Python 2 and py3.4/5
* Increase test coverage and documentation of rank module methods.
* create utils submodule and move util and latent_algebra
* Rename y everywhere to true_labels, and p(true_label=..)
* Enforce positional arguments in methods. Fully remove py2 support.
This was duplicating content in setup.py (and the two had gotten out of
sync, listing different versions for scipy).
According to the Python Packaging User Guide
(https://packaging.python.org/en/latest/discussions/install-requires-vs-requirements/),
the `install_requires` specifies what a project minimally needs to run
correctly, while the requirements.txt file contains an exhaustive list
of pinned versions for the purpose of repeatable installations of a
complete environment. Cleanlab is a Python package that users will `pip
install`, so we don't need a requirements.txt.
Some related projects like scikit-learn don't have a requirements.txt,
and others that do have such a file use it for a different purpose, e.g.
PyTorch and Keras use the file to list dev dependencies.
When multiple coverage reports are uploaded to Codecov, it merges the
reports, which is what we want in this situation. So for example, if the
code branches on Python version or OS version, we'll cover those
branches across tests.
This patch switches the CI from Travis-CI to GitHub Actions.
The new CI script drops support for Python 3.4 (EOLed 18 March 2019) and
Python 3.5 (EOLed 13 September 2020), though it keeps support for Python
2.7 (EOLed 1 January 2020). This keeps the CI configuration a lot
simpler.
The new CI script also simplifies code coverage: a single coverage
report is produced for Python 3.9 on Ubuntu. There are no coverage
statistics being collected and uploaded for older versions of Python 3,
Python 2, or other OSes.
Anish Athalye