Skip to content

Scheduled weekly dependency update for week 39 #627

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Closed
pyup-bot wants to merge 11 commits into from
Closed

Scheduled weekly dependency update for week 39 #627

pyup-bot wants to merge 11 commits into from

Conversation

@pyup-bot
Copy link
Collaborator

@pyup-bot pyup-bot commented Sep 29, 2025

Update sphinx from 7.3.7 to 8.3.0.

Changelog

8.3.0

==============================

Dependencies
------------

Incompatible changes
--------------------

Deprecated
----------

Features added
--------------

Bugs fixed
----------

Testing
-------
Links

Update furo from 2024.5.6 to 2025.9.25.

The bot wasn't able to find a changelog for this release. Got an idea?

Links

Update sphinx-autobuild from 2024.4.16 to 2025.8.25.

Changelog

2025.08.25

-----------------------

* Drop support for Python 3.9-3.10 to match Sphinx.
* Declare support for Python 3.14.
* Add an optional ``--post-build`` flag to run additional commands
after building the documentation.
* Add support for the ``SPHINX_AUTOBUILD_DEBUG`` environment variable
to help with debugging.

2024.10.03

-----------------------

* Improve error handling for failures in pre-build commands.

2024.10.02

-----------------------

* Show the changed paths that triggered the rebuild.

2024.09.19

-----------------------

* Fix path filtering on Windows by normalising path separators.
* Filter various directories by default (``.git``, ``venv``, etc).
* Serve the correct directory when using make mode (``-M``).

2024.09.18

-----------------------

* Run Sphinx through the Python entry point rather than the binary on PATH.

2024.09.17

-----------------------

* Relax checks for paths that aren't required to exist.

2024.09.03

-----------------------

* Fix support for Python 3.9.
* Fix running ``sphinx-autobuild`` via entry point scripts.
* Run ``sphinx-build`` in a subprocess to mitigate autdoc issues.
* Support the ``-M`` 'make mode' option for ``sphinx-build``.
Links

Update sphinx-sitemap from 2.6.0 to 2.8.0.

Changelog

2.8.0

-----

- |:sparkles:| NEW: Add support for wildcard patterns to :confval:`sitemap_excludes`
`113 <https://github.com/jdillard/sphinx-sitemap/pull/113>`_

2.7.2

-----

*Release date: 2025-06-26*

- |:bug:| FIX: Change :confval:`sitemap_show_lastmod` to default of ``False``

2.7.1

-----

*Release date: 2025-06-20*

- Remove support for Python 3.8

2.7.0

-----

*Release date: 2025-06-20*

* |:sparkles:| NEW: Add support for ``lastmod`` using `sphinx-last-updated-by-git`_
`95 <https://github.com/jdillard/sphinx-sitemap/pull/95>`_
Links

Update sphinx-click from 5.0.1 to 6.1.0.

The bot wasn't able to find a changelog for this release. Got an idea?

Links

Update sphinx-autodoc-typehints from 2.3.0 to 3.2.0.

Changelog

3.2.0

<!-- Release notes generated using configuration in .github/release.yml at main -->

What's Changed
* Fixed broken changelog link by agronholm in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/530
* Fix issue 481 by nineteendo in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/533
* Don't add :rtype: None by grayjk in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/538
* Place rtype after directive by grayjk in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/537

New Contributors
* agronholm made their first contribution in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/530
* nineteendo made their first contribution in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/533
* grayjk made their first contribution in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/538

**Full Changelog**: https://github.com/tox-dev/sphinx-autodoc-typehints/compare/3.1.0...3.2.0

3.1.0

<!-- Release notes generated using configuration in .github/release.yml at main -->

What's Changed
* Support Sphinx 8.2.0 - drop 3.10 support because Sphinx does by b-kamphorst in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/525

New Contributors
* b-kamphorst made their first contribution in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/525

**Full Changelog**: https://github.com/tox-dev/sphinx-autodoc-typehints/compare/3.0.1...3.1.0

3.0.1

<!-- Release notes generated using configuration in .github/release.yml at main -->

What's Changed
* Fix roles for `types` module by flying-sheep in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/516
* Update Ruff to 0.9.1 by flying-sheep in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/517


**Full Changelog**: https://github.com/tox-dev/sphinx-autodoc-typehints/compare/3.0.0...3.0.1

3.0.0

<!-- Release notes generated using configuration in .github/release.yml at main -->

What's Changed
* 🐛 Replace docutils private API use w/ public by webknjaz in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/507
* Drop support for nptyping by gaborbernat in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/514

New Contributors
* webknjaz made their first contribution in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/507

**Full Changelog**: https://github.com/tox-dev/sphinx-autodoc-typehints/compare/2.5.0...3.0.0

2.5.0

<!-- Release notes generated using configuration in .github/release.yml at main -->

What's Changed
* Fix the type checker by gaborbernat in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/493
* Apply typehints_formatter to signature by Priyansh121096 in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/494

New Contributors
* Priyansh121096 made their first contribution in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/494

**Full Changelog**: https://github.com/tox-dev/sphinx-autodoc-typehints/compare/2.4.4...2.5.0

2.4.4

<!-- Release notes generated using configuration in .github/release.yml at main -->



**Full Changelog**: https://github.com/tox-dev/sphinx-autodoc-typehints/compare/2.4.3...2.4.4

2.4.3

<!-- Release notes generated using configuration in .github/release.yml at main -->



**Full Changelog**: https://github.com/tox-dev/sphinx-autodoc-typehints/compare/2.4.2...2.4.3

2.4.2

<!-- Release notes generated using configuration in .github/release.yml at main -->

What's Changed
* Improve the CI by gaborbernat in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/485


**Full Changelog**: https://github.com/tox-dev/sphinx-autodoc-typehints/compare/2.4.1...2.4.2

2.4.1

<!-- Release notes generated using configuration in .github/release.yml at main -->

What's Changed
* Fix placement of return type when there is a doctest by hoodmane in https://github.com/tox-dev/sphinx-autodoc-typehints/pull/482


**Full Changelog**: https://github.com/tox-dev/sphinx-autodoc-typehints/compare/2.4.0...2.4.1

2.4.0

<!-- Release notes generated using configuration in .github/release.yml at main -->

- Drop 3.9 support
- Add 3.13 support
- Bump dependencies

**Full Changelog**: https://github.com/tox-dev/sphinx-autodoc-typehints/compare/2.3.0...2.4.0
Links

Update pytest from 8.3.5 to 8.4.2.

Changelog

8.4.1

=========================

Bug fixes
---------

- `13461 <https://github.com/pytest-dev/pytest/issues/13461>`_: Corrected ``_pytest.terminal.TerminalReporter.isatty`` to support
being called as a method. Before it was just a boolean which could
break correct code when using ``-o log_cli=true``).


- `13477 <https://github.com/pytest-dev/pytest/issues/13477>`_: Reintroduced :class:`pytest.PytestReturnNotNoneWarning` which was removed by accident in pytest `8.4`.

This warning is raised when a test functions returns a value other than ``None``, which is often a mistake made by beginners.

See :ref:`return-not-none` for more information.


- `13497 <https://github.com/pytest-dev/pytest/issues/13497>`_: Fixed compatibility with ``Twisted 25+``.



Improved documentation
----------------------

- `13492 <https://github.com/pytest-dev/pytest/issues/13492>`_: Fixed outdated warning about ``faulthandler`` not working on Windows.

8.4.0

=========================

Removals and backward incompatible breaking changes
---------------------------------------------------

- `11372 <https://github.com/pytest-dev/pytest/issues/11372>`_: Async tests will now fail, instead of warning+skipping, if you don't have any suitable plugin installed.


- `12346 <https://github.com/pytest-dev/pytest/issues/12346>`_: Tests will now fail, instead of raising a warning, if they return any value other than None.


- `12874 <https://github.com/pytest-dev/pytest/issues/12874>`_: We dropped support for Python 3.8 following its end of life (2024-10-07).


- `12960 <https://github.com/pytest-dev/pytest/issues/12960>`_: Test functions containing a yield now cause an explicit error. They have not been run since pytest 4.0, and were previously marked as an expected failure and deprecation warning.

See :ref:`the docs <yield tests deprecated>` for more information.



Deprecations (removal in next major release)
--------------------------------------------

- `10839 <https://github.com/pytest-dev/pytest/issues/10839>`_: Requesting an asynchronous fixture without a `pytest_fixture_setup` hook that resolves it will now give a DeprecationWarning. This most commonly happens if a sync test requests an async fixture. This should have no effect on a majority of users with async tests or fixtures using async pytest plugins, but may affect non-standard hook setups or ``autouse=True``. For guidance on how to work around this warning see :ref:`sync-test-async-fixture`.



New features
------------

- `11538 <https://github.com/pytest-dev/pytest/issues/11538>`_: Added :class:`pytest.RaisesGroup` as an equivalent to :func:`pytest.raises` for expecting :exc:`ExceptionGroup`. Also adds :class:`pytest.RaisesExc` which is now the logic behind :func:`pytest.raises` and used as parameter to :class:`pytest.RaisesGroup`. ``RaisesGroup`` includes the ability to specify multiple different expected exceptions, the structure of nested exception groups, and flags for emulating :ref:`except* <except_star>`. See :ref:`assert-matching-exception-groups` and docstrings for more information.


- `12081 <https://github.com/pytest-dev/pytest/issues/12081>`_: Added :fixture:`capteesys` to capture AND pass output to next handler set by ``--capture=``.


- `12504 <https://github.com/pytest-dev/pytest/issues/12504>`_: :func:`pytest.mark.xfail` now accepts :class:`pytest.RaisesGroup` for the ``raises`` parameter when you expect an exception group. You can also pass a :class:`pytest.RaisesExc` if you e.g. want to make use of the ``check`` parameter.


- `12713 <https://github.com/pytest-dev/pytest/issues/12713>`_: New `--force-short-summary` option to force condensed summary output regardless of verbosity level.

This lets users still see condensed summary output of failures for quick reference in log files from job outputs, being especially useful if non-condensed output is very verbose.


- `12749 <https://github.com/pytest-dev/pytest/issues/12749>`_: pytest traditionally collects classes/functions in the test module namespace even if they are imported from another file.

For example:

.. code-block:: python

    contents of src/domain.py
   class Testament: ...


    contents of tests/test_testament.py
   from domain import Testament


   def test_testament(): ...

In this scenario with the default options, pytest will collect the class `Testament` from `tests/test_testament.py` because it starts with `Test`, even though in this case it is a production class being imported in the test module namespace.

This behavior can now be prevented by setting the new :confval:`collect_imported_tests` configuration option to ``false``, which will make pytest collect classes/functions from test files **only** if they are defined in that file.

-- by :user:`FreerGit`


- `12765 <https://github.com/pytest-dev/pytest/issues/12765>`_: Thresholds to trigger snippet truncation can now be set with :confval:`truncation_limit_lines` and :confval:`truncation_limit_chars`.

See :ref:`truncation-params` for more information.


- `13125 <https://github.com/pytest-dev/pytest/issues/13125>`_: :confval:`console_output_style` now supports ``times`` to show execution time of each test.


- `13192 <https://github.com/pytest-dev/pytest/issues/13192>`_: :func:`pytest.raises` will now raise a warning when passing an empty string to ``match``, as this will match against any value. Use ``match="^$"`` if you want to check that an exception has no message.


- `13192 <https://github.com/pytest-dev/pytest/issues/13192>`_: :func:`pytest.raises` will now print a helpful string diff if matching fails and the match parameter has ``^`` and ``$`` and is otherwise escaped.


- `13192 <https://github.com/pytest-dev/pytest/issues/13192>`_: You can now pass :func:`with pytest.raises(check=fn): <pytest.raises>`, where ``fn`` is a function which takes a raised exception and returns a boolean. The ``raises`` fails if no exception was raised (as usual), passes if an exception is raised and ``fn`` returns ``True`` (as well as ``match`` and the type matching, if specified, which are checked before), and propagates the exception if ``fn`` returns ``False`` (which likely also fails the test).


- `13228 <https://github.com/pytest-dev/pytest/issues/13228>`_: :ref:`hidden-param` can now be used in ``id`` of :func:`pytest.param` or in
``ids`` of :py:func:`Metafunc.parametrize <pytest.Metafunc.parametrize>`.
It hides the parameter set from the test name.


- `13253 <https://github.com/pytest-dev/pytest/issues/13253>`_: New flag: :ref:`--disable-plugin-autoload <disable_plugin_autoload>` which works as an alternative to :envvar:`PYTEST_DISABLE_PLUGIN_AUTOLOAD` when setting environment variables is inconvenient; and allows setting it in config files with :confval:`addopts`.



Improvements in existing functionality
--------------------------------------

- `10224 <https://github.com/pytest-dev/pytest/issues/10224>`_: pytest's ``short`` and ``long`` traceback styles (:ref:`how-to-modifying-python-tb-printing`)
now have partial :pep:`657` support and will show specific code segments in the
traceback.

.. code-block:: pytest

   ================================= FAILURES =================================
   _______________________ test_gets_correct_tracebacks _______________________

   test_tracebacks.py:12: in test_gets_correct_tracebacks
       assert manhattan_distance(p1, p2) == 1
              ^^^^^^^^^^^^^^^^^^^^^^^^^^
   test_tracebacks.py:6: in manhattan_distance
       return abs(point_1.x - point_2.x) + abs(point_1.y - point_2.y)
                              ^^^^^^^^^
   E   AttributeError: 'NoneType' object has no attribute 'x'

-- by :user:`ammaraskar`


- `11118 <https://github.com/pytest-dev/pytest/issues/11118>`_: Now :confval:`pythonpath` configures `$PYTHONPATH` earlier than before during the initialization process, which now also affects plugins loaded via the `-p` command-line option.

-- by :user:`millerdev`


- `11381 <https://github.com/pytest-dev/pytest/issues/11381>`_: The ``type`` parameter of the ``parser.addini`` method now accepts `"int"` and ``"float"`` parameters, facilitating the parsing of configuration values in the configuration file.

Example:

.. code-block:: python

 def pytest_addoption(parser):
     parser.addini("int_value", type="int", default=2, help="my int value")
     parser.addini("float_value", type="float", default=4.2, help="my float value")

The `pytest.ini` file:

.. code-block:: ini

 [pytest]
 int_value = 3
 float_value = 5.4


- `11525 <https://github.com/pytest-dev/pytest/issues/11525>`_: Fixtures are now clearly represented in the output as a "fixture object", not as a normal function as before, making it easy for beginners to catch mistakes such as referencing a fixture declared in the same module but not requested in the test function.

-- by :user:`the-compiler` and :user:`glyphack`


- `12426 <https://github.com/pytest-dev/pytest/issues/12426>`_: A warning is now issued when :ref:`pytest.mark.usefixtures ref` is used without specifying any fixtures. Previously, empty usefixtures markers were silently ignored.


- `12707 <https://github.com/pytest-dev/pytest/issues/12707>`_: Exception chains can be navigated when dropped into Pdb in Python 3.13+.


- `12736 <https://github.com/pytest-dev/pytest/issues/12736>`_: Added a new attribute `name` with the fixed value `"pytest tests"` to the root tag `testsuites` of the junit-xml generated by pytest.

This attribute is part of many junit-xml specifications and is even part of the `junit-10.xsd` specification that pytest's implementation is based on.


- `12943 <https://github.com/pytest-dev/pytest/issues/12943>`_: If a test fails with an exceptiongroup with a single exception, the contained exception will now be displayed in the short test summary info.


- `12958 <https://github.com/pytest-dev/pytest/issues/12958>`_: A number of :ref:`unraisable <unraisable>` enhancements:

* Set the unraisable hook as early as possible and unset it as late as possible, to collect the most possible number of unraisable exceptions.
* Call the garbage collector just before unsetting the unraisable hook, to collect any straggling exceptions.
* Collect multiple unraisable exceptions per test phase.
* Report the :mod:`tracemalloc` allocation traceback (if available).
* Avoid using a generator based hook to allow handling :class:`StopIteration` in test failures.
* Report the unraisable exception as the cause of the :class:`pytest.PytestUnraisableExceptionWarning` exception if raised.
* Compute the ``repr`` of the unraisable object in the unraisable hook so you get the latest information if available, and should help with resurrection of the object.


- `13010 <https://github.com/pytest-dev/pytest/issues/13010>`_: :func:`pytest.approx` now can compare collections that contain numbers and non-numbers mixed.


- `13016 <https://github.com/pytest-dev/pytest/issues/13016>`_: A number of :ref:`threadexception <unraisable>` enhancements:

* Set the excepthook as early as possible and unset it as late as possible, to collect the most possible number of unhandled exceptions from threads.
* Collect multiple thread exceptions per test phase.
* Report the :mod:`tracemalloc` allocation traceback (if available).
* Avoid using a generator based hook to allow handling :class:`StopIteration` in test failures.
* Report the thread exception as the cause of the :class:`pytest.PytestUnhandledThreadExceptionWarning` exception if raised.
* Extract the ``name`` of the thread object in the excepthook which should help with resurrection of the thread.


- `13031 <https://github.com/pytest-dev/pytest/issues/13031>`_: An empty parameter set as in ``pytest.mark.parametrize([], ids=idfunc)`` will no longer trigger a call to ``idfunc`` with internal objects.


- `13115 <https://github.com/pytest-dev/pytest/issues/13115>`_: Allows supplying ``ExceptionGroup[Exception]`` and ``BaseExceptionGroup[BaseException]`` to ``pytest.raises`` to keep full typing on :class:`ExceptionInfo <pytest.ExceptionInfo>`:

.. code-block:: python

   with pytest.raises(ExceptionGroup[Exception]) as exc_info:
       some_function()

Parametrizing with other exception types remains an error - we do not check the types of child exceptions and thus do not permit code that might look like we do.


- `13122 <https://github.com/pytest-dev/pytest/issues/13122>`_: The ``--stepwise`` mode received a number of improvements:

* It no longer forgets the last failed test in case pytest is executed later without the flag.

 This enables the following workflow:

 1. Execute pytest with ``--stepwise``, pytest then stops at the first failing test;
 2. Iteratively update the code and run the test in isolation, without the ``--stepwise`` flag (for example in an IDE), until it is fixed.
 3. Execute pytest with ``--stepwise`` again and pytest will continue from the previously failed test, and if it passes, continue on to the next tests.

 Previously, at step 3, pytest would start from the beginning, forgetting the previously failed test.

 This change however might cause issues if the ``--stepwise`` mode is used far apart in time, as the state might get stale, so the internal state will be reset automatically in case the test suite changes (for now only the number of tests are considered for this, we might change/improve this on the future).

* New ``--stepwise-reset``/``--sw-reset`` flag, allowing the user to explicitly reset the stepwise state and restart the workflow from the beginning.


- `13308 <https://github.com/pytest-dev/pytest/issues/13308>`_: Added official support for Python 3.14.


- `13380 <https://github.com/pytest-dev/pytest/issues/13380>`_: Fix :class:`ExceptionGroup` traceback filtering to exclude pytest internals.


- `13415 <https://github.com/pytest-dev/pytest/issues/13415>`_: The author metadata of the BibTex example is now correctly formatted with last names following first names.
An example of BibLaTex has been added.
BibTex and BibLaTex examples now clearly indicate that what is cited is software.

-- by :user:`willynilly`


- `13420 <https://github.com/pytest-dev/pytest/issues/13420>`_: Improved test collection performance by optimizing path resolution used in ``FSCollector``.


- `13457 <https://github.com/pytest-dev/pytest/issues/13457>`_: The error message about duplicate parametrization no longer displays an internal stack trace.


- `4112 <https://github.com/pytest-dev/pytest/issues/4112>`_: Using :ref:`pytest.mark.usefixtures <pytest.mark.usefixtures ref>` on :func:`pytest.param` now produces an error instead of silently doing nothing.


- `5473 <https://github.com/pytest-dev/pytest/issues/5473>`_: Replace `:` with `;` in the assertion rewrite warning message so it can be filtered using standard Python warning filters before calling :func:`pytest.main`.


- `6985 <https://github.com/pytest-dev/pytest/issues/6985>`_: Improved :func:`pytest.approx` to enhance the readability of value ranges and tolerances between 0.001 and 1000.
 * The `repr` method now provides clearer output for values within those ranges, making it easier to interpret the results.
 * Previously, the output for those ranges of values and tolerances was displayed in scientific notation (e.g., `42 ± 1.0e+00`). The updated method now presents the tolerance as a decimal for better readability (e.g., `42 ± 1`).

   Example:

   **Previous Output:**

   .. code-block:: console

       >>> pytest.approx(42, abs=1)
       42 ± 1.0e+00

   **Current Output:**

   .. code-block:: console

       >>> pytest.approx(42, abs=1)
       42 ± 1

 -- by :user:`fazeelghafoor`


- `7683 <https://github.com/pytest-dev/pytest/issues/7683>`_: The formerly optional ``pygments`` dependency is now required, causing output always to be source-highlighted (unless disabled via the ``--code-highlight=no`` CLI option).



Bug fixes
---------

- `10404 <https://github.com/pytest-dev/pytest/issues/10404>`_: Apply filterwarnings from config/cli as soon as possible, and revert them as late as possible
so that warnings as errors are collected throughout the pytest run and before the
unraisable and threadexcept hooks are removed.

This allows very late warnings and unraisable/threadexcept exceptions to fail the test suite.

This also changes the warning that the lsof plugin issues from PytestWarning to the new warning PytestFDWarning so it can be more easily filtered.


- `11067 <https://github.com/pytest-dev/pytest/issues/11067>`_: The test report is now consistent regardless if the test xfailed via :ref:`pytest.mark.xfail <pytest.mark.xfail ref>` or :func:`pytest.fail`.

Previously, *xfailed* tests via the marker would have the string ``"reason: "`` prefixed to the message, while those *xfailed* via the function did not. The prefix has been removed.


- `12008 <https://github.com/pytest-dev/pytest/issues/12008>`_: In :pr:`11220`, an unintended change in reordering was introduced by changing the way indices were assigned to direct params. More specifically, before that change, the indices of direct params to metafunc's callspecs were assigned after all parametrizations took place. Now, that change is reverted.


- `12863 <https://github.com/pytest-dev/pytest/issues/12863>`_: Fix applying markers, including :ref:`pytest.mark.parametrize <pytest.mark.parametrize ref>` when placed above `staticmethod` or `classmethod`.


- `12929 <https://github.com/pytest-dev/pytest/issues/12929>`_: Handle StopIteration from test cases, setup and teardown correctly.


- `12938 <https://github.com/pytest-dev/pytest/issues/12938>`_: Fixed ``--durations-min`` argument not respected if ``-vv`` is used.


- `12946 <https://github.com/pytest-dev/pytest/issues/12946>`_: Fixed missing help for :mod:`pdb` commands wrapped by pytest -- by :user:`adamchainz`.


- `12981 <https://github.com/pytest-dev/pytest/issues/12981>`_: Prevent exceptions in :func:`pytest.Config.add_cleanup` callbacks preventing further cleanups.


- `13047 <https://github.com/pytest-dev/pytest/issues/13047>`_: Restore :func:`pytest.approx` handling of equality checks between `bool` and `numpy.bool_` types.

Comparing `bool` and `numpy.bool_` using :func:`pytest.approx` accidentally changed in version `8.3.4` and `8.3.5` to no longer match:

.. code-block:: pycon

   >>> import numpy as np
   >>> from pytest import approx
   >>> [np.True_, np.True_] == pytest.approx([True, True])
   False

This has now been fixed:

.. code-block:: pycon

   >>> [np.True_, np.True_] == pytest.approx([True, True])
   True


- `13119 <https://github.com/pytest-dev/pytest/issues/13119>`_: Improved handling of invalid regex patterns for filter warnings by providing a clear error message.


- `13175 <https://github.com/pytest-dev/pytest/issues/13175>`_: The diff is now also highlighted correctly when comparing two strings.


- `13248 <https://github.com/pytest-dev/pytest/issues/13248>`_: Fixed an issue where passing a ``scope`` in :py:func:`Metafunc.parametrize <pytest.Metafunc.parametrize>` with ``indirect=True``
could result in other fixtures being unable to depend on the parametrized fixture.


- `13291 <https://github.com/pytest-dev/pytest/issues/13291>`_: Fixed ``repr`` of ``attrs`` objects in assertion failure messages when using ``attrs>=25.2``.


- `13312 <https://github.com/pytest-dev/pytest/issues/13312>`_: Fixed a possible ``KeyError`` crash on PyPy during collection of tests involving higher-scoped parameters.


- `13345 <https://github.com/pytest-dev/pytest/issues/13345>`_: Fix type hints for :attr:`pytest.TestReport.when` and :attr:`pytest.TestReport.location`.


- `13377 <https://github.com/pytest-dev/pytest/issues/13377>`_: Fixed handling of test methods with positional-only parameter syntax.

Now, methods are supported that formally define ``self`` as positional-only
and/or fixture parameters as keyword-only, e.g.:

.. code-block:: python

   class TestClass:

       def test_method(self, /, *, fixture): ...

Before, this caused an internal error in pytest.


- `13384 <https://github.com/pytest-dev/pytest/issues/13384>`_: Fixed an issue where pytest could report negative durations.


- `13420 <https://github.com/pytest-dev/pytest/issues/13420>`_: Added ``lru_cache`` to ``nodes._check_initialpaths_for_relpath``.


- `9037 <https://github.com/pytest-dev/pytest/issues/9037>`_: Honor :confval:`disable_test_id_escaping_and_forfeit_all_rights_to_community_support` when escaping ids in parametrized tests.



Improved documentation
----------------------

- `12535 <https://github.com/pytest-dev/pytest/issues/12535>`_: `This
example`<https://docs.pytest.org/en/latest/example/simple.html#making-test-result-information-available-in-fixtures>
showed ``print`` statements that do not exactly reflect what the
different branches actually do.  The fix makes the example more precise.


- `13218 <https://github.com/pytest-dev/pytest/issues/13218>`_: Pointed out in the :func:`pytest.approx` documentation that it considers booleans unequal to numeric zero or one.


- `13221 <https://github.com/pytest-dev/pytest/issues/13221>`_: Improved grouping of CLI options in the ``--help`` output.


- `6649 <https://github.com/pytest-dev/pytest/issues/6649>`_: Added :class:`~pytest.TerminalReporter` to the :ref:`api-reference` documentation page.


- `8612 <https://github.com/pytest-dev/pytest/issues/8612>`_: Add a recipe for handling abstract test classes in the documentation.

A new example has been added to the documentation to demonstrate how to use a mixin class to handle abstract
test classes without manually setting the ``__test__`` attribute for subclasses.
This ensures that subclasses of abstract test classes are automatically collected by pytest.



Packaging updates and notes for downstreams
-------------------------------------------

- `13317 <https://github.com/pytest-dev/pytest/issues/13317>`_: Specified minimum allowed versions of ``colorama``, ``iniconfig``,
and ``packaging``; and bumped the minimum allowed version
of ``exceptiongroup`` for ``python_version<'3.11'`` from a release
candidate to a full release.



Contributor-facing changes
--------------------------

- `12017 <https://github.com/pytest-dev/pytest/issues/12017>`_: Mixed internal improvements:

* Migrate formatting to f-strings in some tests.
* Use type-safe constructs in JUnitXML tests.
* Moved`` MockTiming`` into ``_pytest.timing``.

-- by :user:`RonnyPfannschmidt`


- `12647 <https://github.com/pytest-dev/pytest/issues/12647>`_: Fixed running the test suite with the ``hypothesis`` pytest plugin.



Miscellaneous internal changes
------------------------------

- `6649 <https://github.com/pytest-dev/pytest/issues/6649>`_: Added :class:`~pytest.TerminalReporter` to the public pytest API, as it is part of the signature of the :hook:`pytest_terminal_summary` hook.
Links

Update pytest-mock from 3.14.0 to 3.15.1.

Changelog

3.15.1

------

*2025-09-16*

* `529 <https://github.com/pytest-dev/pytest-mock/issues/529>`_: Fixed ``itertools._tee object has no attribute error`` -- now ``duplicate_iterators=True`` must be passed to ``mocker.spy`` to duplicate iterators.

3.15.0

------

*2025-09-04*

* Python 3.8 (EOL) is no longer supported.
* `524 <https://github.com/pytest-dev/pytest-mock/pull/524>`_: Added ``spy_return_iter`` to ``mocker.spy``, which contains a duplicate of the return value of the spied method if it is an ``Iterator``.

3.14.1

-------------------

* `503 <https://github.com/pytest-dev/pytest-mock/pull/503>`_: Python 3.14 is now officially supported.
Links

Update fiftyone from 1.5.2 to 1.8.1.

Changelog

1.8.1

Please see the official [Voxel51 Release Notes](https://docs.voxel51.com/release-notes.html#fiftyone-1-8-1) for details.

1.8.0

Please see the official [Voxel51 Release Notes](https://docs.voxel51.com/release-notes.html#fiftyone-1-8-0) for details.

1.7.2

Please see the official [Voxel51 Release Notes](https://docs.voxel51.com/release-notes.html#fiftyone-1-7-2) for details.

1.7.1

Please see the official [Voxel51 Release Notes](https://docs.voxel51.com/release-notes.html#fiftyone-1-7-1) for details.

1.7.0

Please see the official [Voxel51 Release Notes](https://docs.voxel51.com/release-notes.html#fiftyone-1-7-0) for details!

1.6.0

Please see the official [Voxel51 Release Notes](https://docs.voxel51.com/release-notes.html#fiftyone-1-6-0) for details.
Links

Update datasets from 3.6.0 to 4.1.1.

Changelog

4.1.0

Added

-   It is now easier to create datasets outside TFDS repository (see our updated
 [dataset creation guide](https://www.tensorflow.org/datasets/add_dataset)).
-   When generating a dataset, if download fails for any reason, it is now
 possible to manually download the data. See
 [doc](https://www.tensorflow.org/datasets/overview#manual_download_if_download_fails).
-   `tfds.core.as_path` to create pathlib.Path-like objects compatible with GCS
 (e.g. `tfds.core.as_path('gs://my-bucket/labels.csv').read_text()`).
-   `verify_ssl=` option to `tfds.download.DownloadConfig` to disable SSH
 certificate during download.
-   New datasets.  Changed
-   All dataset inherit from `tfds.core.GeneratorBasedBuilder`. Converting a
 dataset to beam now only require changing `_generate_examples` (see
 [example and doc](https://www.tensorflow.org/datasets/beam_datasets#instructions)).
-   `_split_generators` should now returns `{'split_name':
 self._generate_examples(), ...}` (but current datasets are backward
 compatible).
-   Better `pathlib.Path`, `os.PathLike` compatibility: `dl_manager.manual_dir`
 now returns a pathlib-Like object. Example: `python text =
 (dl_manager.manual_dir / 'downloaded-text.txt').read_text()` Note: Other
 `dl_manager.download`, `.extract`,... will return pathlib-like objects in
 future versions. `FeatureConnector`,... and most functions should accept
 `PathLike` objects. Let us know if some functions you need are missing.
-   `--record_checksums` now assume the new dataset-as-folder model.

Deprecated

-   `tfds.core.SplitGenerator`, `tfds.core.BeamBasedBuilder` are deprecated and
 will be removed in a future version.

Fixed

-   `BuilderConfig` are now compatible with Beam datasets 2348
-   `tfds.features.Images` can accept encoded `bytes` images directly (useful
 when used with `img_name, img_bytes =
 dl_manager.iter_archive('images.zip')`).
-   Doc API now show deprecated methods, abstract methods to overwrite are now
 documented.
-   You can generate `imagenet2012` with only a single split (e.g. only the
 validation data). Other split will be skipped if not present.

4.0.1

Fixed

-   `tfds.load` when generation code isn't present.
-   GCS compatibility.

4.0.0

Added

-   Dataset-as-folder: Dataset can now be self-contained module in a folder with
 checksums, dummy data,... This simplify implementing datasets outside the
 TFDS repository.
-   `tfds.load` can now load dataset without using the generation class. So
 `tfds.load('my_dataset:1.0.0')` can work even if `MyDataset.VERSION ==
 '2.0.0'` (See 2493).
-   TFDS CLI (see https://www.tensorflow.org/datasets/cli for detail).
-   `tfds.testing.mock_data` does not require metadata files anymore!
-   `tfds.as_dataframe(ds, ds_info)` with custom visualisation
 ([example](https://www.tensorflow.org/datasets/overview#tfdsas_dataframe)).
-   `tfds.even_splits` to generate subsplits (e.g. `tfds.even_splits('train',
 n=3) == ['train[0%:33%]', 'train[33%:67%]', ...]`.
-   `DatasetBuilder.RELEASE_NOTES` property.
-   `tfds.features.Image` now supports PNG with 4-channels.
-   `tfds.ImageFolder` now supports custom shape, dtype.
-   Downloaded URLs are available through `MyDataset.url_infos`.
-   `skip_prefetch` option to `tfds.ReadConfig`.
-   `as_supervised=True` support for `tfds.show_examples`, `tfds.as_dataframe`.
-   tfds.features can now be saved/loaded, you may have to overwrite
 [FeatureConnector.from_json_content](https://www.tensorflow.org/datasets/api_docs/python/tfds/features/FeatureConnector?version=nightly#from_json_content)
 and `FeatureConnector.to_json_content` to support this feature.
-   Script to detect dead-urls.
-   New datasets.

Changed

-   `tfds.as_numpy()` now returns an iterable which can be iterated multiple
 times. To migrate: `next(ds)` -> `next(iter(ds))`.
-   Rename `tfds.features.text.Xyz` -> `tfds.deprecated.text.Xyz`.

Removed

-   `DatasetBuilder.IN_DEVELOPMENT` property.
-   `tfds.core.disallow_positional_args` (should use Py3 `*,` instead).
-   Testing against TF 1.15. Requires Python 3.6.8+.

Fixed

-   Better archive extension detection for `dl_manager.download_and_extract`.
-   Fix `tfds.__version__` in TFDS nightly to be PEP440 compliant
-   Fix crash when GCS not available.
-   Improved open-source workflow, contributor guide, documentation.
-   Many other internal cleanups, bugs, dead code removal, py2->py3 cleanup,
 pytype annotations,...
-   Datasets updates.
Links

Update ultralytics from 8.3.132 to 8.3.203.

Changelog

8.3.203

🌟 Summary
Safer, smoother training resumes with AMP plus sturdier exports and better version checks. This release improves mixed‑precision resume stability, future‑proofs ONNX opset selection, tightens export requirements (OpenVINO/CoreML), and polishes compatibility across models and PyTorch versions. 🚀🛡️

📊 Key Changes
- Training resume reliability (priority)
- Restore AMP GradScaler when resuming training to keep loss scaling stable [PR 22189](https://github.com/ultralytics/ultralytics/pull/22189) (fixes [#21913](https://github.com/ultralytics/ultralytics/issues/21913), [#17282](https://github.com/ultralytics/ultralytics/issues/17282)) ⚖️
- ONNX export robustness
- Smart opset selection via `best_onnx_opset()` and IR version cap for better ONNX Runtime compatibility [PR 22142](https://github.com/ultralytics/ultralytics/pull/22142) 🔧
- Clear Torch gating for opset logic (Torch ≥ 1.13 uses ONNX_MAX_OPSET-1; older Torch uses a map) [PR 22155](https://github.com/ultralytics/ultralytics/pull/22155) 🧭
- Export requirements updates
- OpenVINO export now requires PyTorch ≥ 2.1 [PR 22156](https://github.com/ultralytics/ultralytics/pull/22156) ⚠️
- CoreML export now requires PyTorch ≥ 1.11 [PR 22162](https://github.com/ultralytics/ultralytics/pull/22162) 🍎
- Model and training flow improvements
- Disable `compile=True` for YOLO-World and YOLOE with clear assertions; keep compile-optimized path when enabled elsewhere [PR 22148](https://github.com/ultralytics/ultralytics/pull/22148) 🧠
- RTDETR requires PyTorch ≥ 1.11 with runtime asserts and aligned tests [PR 22161](https://github.com/ultralytics/ultralytics/pull/22161) 🎯
- Data and inference fixes
- Correctly detect image URLs with query strings (no infinite stream loop) [PR 22122](https://github.com/ultralytics/ultralytics/pull/22122) 🌐✅
- Compatibility cleanups
- Unified meshgrid version flags to avoid errors on older Torch versions [PR 22158](https://github.com/ultralytics/ultralytics/pull/22158), [PR #22163](https://github.com/ultralytics/ultralytics/pull/22163) 🧩
- FastSAM workaround for PyTorch 1.9 mask summation bug [PR 22169](https://github.com/ultralytics/ultralytics/pull/22169) 🩹
- Reliability and UX
- Warn if multiple Ultralytics installations are detected to prevent conflicts [PR 22131](https://github.com/ultralytics/ultralytics/pull/22131) 🔍
- Safer tuner cleanup by checking `best_save_dir` exists [PR 22137](https://github.com/ultralytics/ultralytics/pull/22137) 🧹
- Docs and tooling
- Added reference docs for `best_onnx_opset` and multiple-install checks; improved slug generation for content tabs; corrected CI badge; enhanced Model YAML Guide and nav [PRs 22142](https://github.com/ultralytics/ultralytics/pull/22142), [#22157](https://github.com/ultralytics/ultralytics/pull/22157), [#22184](https://github.com/ultralytics/ultralytics/pull/22184), [#22154](https://github.com/ultralytics/ultralytics/pull/22154) 📚
- Pin dev tooling: uv set to 0.8.19 for consistent environments [PR 22186](https://github.com/ultralytics/ultralytics/pull/22186) 🧰

🎯 Purpose & Impact
- More reliable mixed-precision resumes 🎛️
- Restoring the GradScaler avoids instability and prolonged warm-up when continuing training with AMP.
- Fewer export surprises across environments 📦
- ONNX opset selection and IR capping reduce runtime/load errors.
- Clear, upfront version requirements for OpenVINO (Torch ≥ 2.1) and CoreML (Torch ≥ 1.11) prevent confusing failures.
- Better user guidance and DX 🧭
- Early warnings for multiple installs help resolve version/path conflicts quickly.
- Compile behavior is explicit for YOLO-World and YOLOE, reducing trial-and-error.
- Broader compatibility and stability 🧱
- Meshgrid and FastSAM fixes ensure older Torch users can run without errors.
- Smoother tuning and documentation experience ✨
- Safer cleanup logic and improved docs navigation/references make workflows clearer and more predictable.

Tip: If you export to OpenVINO or CoreML, verify your PyTorch version first:
python
import torch
print(torch.__version__)

Upgrade instructions: see the official PyTorch install guide on the PyTorch website.

What's Changed
* Fix Tuner missing `best_save_dir` by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22137
* Add check to warn about multiple Ultralytics installation by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22131
* Disable `compile` for WorldModel and YOLOEModel by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22148
* Fix url image inference by Burhan-Q in https://github.com/ultralytics/ultralytics/pull/22122
* `ultralytics 8.3.203` Future-proof ONNX `opset` selection by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22142
* Use opset table for `torch<1.13` by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22155
* Require `torch>=2.1` for OpenVINO exports by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22156
* Fix meshgrid with indexing bug on `torch==1.10` by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22158
* Require `torch>=1.13` for YOLOE models by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22160
* Enable slug generation for content tab anchor links in MkDocs by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22157
* Require `torch>=1.11` for RTDETR models by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22161
* Fix transformer.py meshgrid for `torch<1.11` by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22163
* Require `torch>=1.11` for CoreML exports by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22162
* PyTorch 1.9 FastSAM `sum(0).sum()` predictor fix by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22169
* Enable comments and add cloning instruction to Model YAML configuration guide by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22154
* Add Actions CI badge to Docs by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22184
* Add `uv` `required-version = "==0.8.19"` by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22186
* `ultralytics 8.3.203` Restore `GradScaler` on resuming training by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22189


**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.3.202...v8.3.203

8.3.202

🌟 Summary
TFLite export gets a targeted INT8 upgrade: per-channel quantization is fixed and INT8 exports are slimmer and more reliable in ultralytics 8.3.202. 🚀

📊 Key Changes
- TFLite INT8 per-channel quantization fix by Y-T-G, resolving [Issue 21030](https://github.com/ultralytics/ultralytics/issues/21030). 🔧 See PR: [TFLite per-channel INT8 quantization fix](https://github.com/ultralytics/ultralytics/pull/22133)
- Removed explicit per-tensor override so per-channel INT8 works as intended (more accurate for conv weights).
- Disabled batchmatmul_unfold when exporting INT8 to avoid massive integer_quant file sizes and improve deployment practicality.
- Distributed Tuner robustness: if a MongoDB collection exists but is empty, tuning now seeds immediately instead of falling back to CSV. 🤝 See PR: [Start distributed Tuner mutation if Mongo collection exists](https://github.com/ultralytics/ultralytics/pull/22120) by glenn-jocher
- CI stability: macOS runner pinned to macos-26 for consistent builds. 🧪 See PR: [Update CI workflow to run on macOS 26](https://github.com/ultralytics/ultralytics/pull/22119) by onuralpszr

🎯 Purpose & Impact
- Better accuracy and compatibility for TFLite INT8 models: Per-channel quantization is typically more accurate for convolution-heavy models like YOLO11. 📈
- Smaller INT8 artifacts and easier deployment: Disabling batchmatmul_unfold for INT8 prevents integer_quant files from ballooning (e.g., ~27MB vs ~3.2MB), making edge deployments lighter and faster to ship. 📦
- Smoother distributed hyperparameter tuning: Immediate start in multi-worker setups when a MongoDB collection exists, improving efficiency in large experiments. ⚙️
- More reliable CI: Fewer runner surprises, faster iteration for contributors, and steadier releases for users. ✅

Quick tip to export INT8 TFLite with the new behavior:
python
from ultralytics import YOLO

model = YOLO("yolo11n.pt")   or your trained YOLO11 model
model.export(format="tflite", int8=True)   per-channel INT8 fix applied; leaner artifacts


Enjoy improved TFLite quantization and leaner edge-ready models! ✨

What's Changed
* Start distributed Tuner mutation if Mongo collection exists by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22120
* ci: 👷 Update CI workflow to run on macOS 26 by onuralpszr in https://github.com/ultralytics/ultralytics/pull/22119
* `ultralytics 8.3.202` TFLite `per-channel` INT8 quantization fix by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22133


**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.3.201...v8.3.202

8.3.201

📊 Key Changes
- TensorFlow export compatibility expanded (primary change) 🔧
- Relaxed `ai-edge-litert` requirement to `>=1.2.0` for non-macOS platforms; macOS remains pinned to `<1.4.0` to avoid known issues.
- Minor typing improvement: `check_requirements()` now accepts `tuple[str]` as well as `Path | str | list[str]`.  
- See details in the current PR: [Expand TensorFlow ai-edge-litert≥1.2 support (non-macOS)](https://github.com/ultralytics/ultralytics/pull/22110) by glenn-jocher.

- Export robustness and Docker reliability 🐳
- Exporter now validates non-zero file size and correctly logs tuple-returning exporters for accurate success reporting: [Assert non-zero filesize in Exporter](https://github.com/ultralytics/ultralytics/pull/22108) by glenn-jocher.
- Docker export images install TensorRT on-demand (no pre-install) to match runtime CUDA, reducing version conflicts: [Use AutoInstall for TensorRT package](https://github.com/ultralytics/ultralytics/pull/22084) by Y-T-G.
- Dockerfiles export using a local temp path (`tmp/yolo11n.pt`) for EdgeTPU/NCNN/IMX steps, improving stability and reproducibility: [Export in Dockerfile /tmp dir](https://github.com/ultralytics/ultralytics/pull/22107) by glenn-jocher.

- Better CLI help for Ultralytics HUB 🧭
- `yolo -h` now shows the correct context-aware help for HUB subcommands by adjusting help precedence: [Display help message with yolo -h](https://github.com/ultralytics/ultralytics/pull/22102) by Y-T-G.

- Training flexibility for multimodal (YOLOE) workflows 🧠
- Removed training-mode assertion to allow `get_text_pe()` during training: [Remove training mode assertion in get_text_pe()](https://github.com/ultralytics/ultralytics/pull/22076) by Y-T-G.

- Dataset reliability fix 📦
- Correct CSV parsing for SKU-110K (no-header CSVs) to prevent misaligned columns: [Fix CSV read error in SKU-110K.yaml](https://github.com/ultralytics/ultralytics/pull/22094) by Y-T-G.

- Documentation improvements 📚
- New in-depth guide: [Model YAML Configuration Guide](https://github.com/ultralytics/ultralytics/blob/main/docs/en/guides/model-yaml-config.md) — architecture structure, scaling, TorchVision backbones, and custom modules: [Docs PR](https://github.com/ultralytics/ultralytics/pull/21830) by Y-T-G.
- Augmentation docs now show a “Supported Tasks” column to clarify applicability across detect/segment/pose/OBB/classify: [Add Supported Tasks column](https://github.com/ultralytics/ultralytics/pull/21900) by Y-T-G.

🎯 Purpose & Impact
- Smoother TensorFlow export workflows ✅
- Fewer dependency conflicts and better forward-compatibility for ONNX→TensorFlow exports on Linux/Windows.
- macOS stability preserved with a safe upper bound on `ai-edge-litert`. 🛡️

- More reliable and portable Docker builds 🔁
- On-demand TensorRT avoids CUDA/TensorRT mismatches.
- Temp-path exports reduce flakiness and speed up CI/CD builds.

- Clearer, more helpful CLI and logs 🗺️
- `yolo -h` behaves as users expect, especially for Ultralytics HUB commands.
- Exporter now catches empty outputs early and logs the correct paths and sizes.

- Enhanced training flexibility and dataset setup 🧪
- Enables text positional embeddings retrieval during training for multimodal YOLOE scenarios.
- SKU-110K dataset downloads parse cleanly, reducing setup hiccups.

- Easier customization and configuration 📘
- The Model YAML guide and updated augmentation reference make it simpler to design, scale, and deploy custom YOLO11 models with confidence.

Tip: To try TensorFlow SavedModel export with the updated dependency handling:
bash
pip install ultralytics
yolo export model=yolo11n.pt format=saved_model


What's Changed
* Fix CSV read error in SKU-110K.yaml download script by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22094
* Remove training mode assertion in `get_text_pe()` by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22076
* Update Dockerfile-export to use AutoInstall for TensorRT package by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22084
* Export in Dockerfile `/tmp` dir by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22107
* Display help message with `yolo -h` by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22102
* Assert non-zero filesize in Exporter by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22108
* Add "Supported Tasks" column to augmentation arguments table by Y-T-G in https://github.com/ultralytics/ultralytics/pull/21900
* Docs: Add comprehensive model YAML configuration guide by Y-T-G in https://github.com/ultralytics/ultralytics/pull/21830
* `ultralytics 8.3.201` Expand TensorFlow `ai-edge-litert≥1.2` support (non-macOS) by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22110


**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.3.200...v8.3.201

8.3.200

🌟 Summary
A focused export refactor that adds a complete Sony IMX500 pipeline, cleans up IMX-specific logic in model heads, and introduces clearer export APIs—plus stability fixes for ONNX (OBB), TensorRT exports, and smoother multi-GPU training. 🚀

📊 Key Changes
- Export refactor and new IMX pipeline (PR [22005](https://github.com/ultralytics/ultralytics/pull/22005)) ✨
- New, clearer export APIs:
 - `export_onnx` → `torch2onnx`
 - `export_engine` → `onnx2engine`
- New Sony IMX export:
 - `torch2imx` utility with quantization and IMX-friendly NMS wrapping
 - IMX-specific helpers (`FXModel`, Detect/Pose overrides, `NMSWrapper`)
- Export utilities reorganized to `ultralytics/utils/export/` with `imx.py`
- Detect/Pose heads simplified by removing IMX-only branches
- Removed `FXModel` from `torch_utils`
- Docs updated to reflect the new structure

- IMX support clarity (PR [22082](https://github.com/ultralytics/ultralytics/pull/22082)) 🧭
- Error messages now explicitly state IMX exports support YOLOv8n and YOLO11n (detection only in benchmarks).

- ONNX + OBB reliability (PR [22079](https://github.com/ultralytics/ultralytics/pull/22079)) 🧩
- Fixes a RuntimeError for OBB models exported with NMS on CUDA by safeguarding empty-mask cases.

- TensorRT export stability (PR [22080](https://github.com/ultralytics/ultralytics/pull/22080)) ⚙️
- Corrects dynamic shape handling and re-enables TensorRT tests
- Extends GPU CI timeout from 20 → 60 minutes for reliability

- DDP training improvements (PR [22073](https://github.com/ultralytics/ultralytics/pull/22073)) 💼
- Initializes `world_size`/`ddp` earlier, simplifies DDP APIs, and prevents duplicate wandb logs

- Tuner robustness (PR [22066](https://github.com/ultralytics/ultralytics/pull/22066)) 🛡️
- Avoids `KeyError` when `close_mosaic` isn’t in the hyperparameter search space

- UX polish (PRs [22071](https://github.com/ultralytics/ultralytics/pull/22071), [#22083](https://github.com/ultralytics/ultralytics/pull/22083), [#22046](https://github.com/ultralytics/ultralytics/pull/22046), [#22063](https://github.com/ultralytics/ultralytics/pull/22063), [#22085](https://github.com/ultralytics/ultralytics/pull/22085)) ✍️
- Aligns multi-GPU device printing
- Clearer export error when trying to “export” to `pt`
- Fixes a broken DeepStream link for Jetson users
- Documentation typo and formatting cleanups
- Updates docs authors metadata

🎯 Purpose & Impact
- Cleaner, more intuitive export APIs 🧭
- The `torch2onnx` and `onnx2engine` names make conversion stages explicit and reduce confusion.
- First-class Sony IMX exports 📦
- New `torch2imx` enables end-to-end IMX500 deployment with built-in quantization and NMS wrapping.
- Clearer docs and package layout make maintenance and contributions easier.
- More reliable exports across backends ✅
- ONNX OBB fix prevents runtime failures on CUDA.
- TensorRT export is more dependable, with re-enabled coverage to catch regressions earlier.
- Smoother multi-GPU training 🧠
- Early DDP setup avoids double logging and simplifies internals, improving stability for distributed runs.
- Fewer paper cuts for users ✂️
- Robust tuner behavior, better error messages, aligned device output, and updated docs links all reduce friction.

Quick tips:
- Export to ONNX:
- CLI: `yolo export model=yolo11n.pt format=onnx`
- Export to TensorRT:
- CLI: `yolo export model=yolo11n.pt format=engine dynamic`
- Export to Sony IMX:
- CLI: `yolo export model=yolo11n.pt format=imx`
- Note: IMX benchmarks currently enforce detection-only and support YOLOv8n/YOLO11n.

What's Changed
* Fix `KeyError` when `close_mosaic` not part of search space by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22066
* Update yolo-data-augmentation.md by miles-deans-ultralytics in https://github.com/ultralytics/ultralytics/pull/22063
* Update deepstream-nvidia-jetson.md by miles-deans-ultralytics in https://github.com/ultralytics/ultralytics/pull/22046
* Revert "Re-enable TensorRT export in GPU tests" by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22078
* Fix DDP device spacing by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22071
* Add Onuralp and Miles to MkDocs GitHub authors by glenn-jocher in https://github.com/ultralytics/ultralytics/pull/22085
* Improve export error message for `pt` format by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/22083
* Update `IMX` export assert message to include `YOLO11n` by RizwanMunawar in https://github.com/ultralytics/ultralytics/pull/22082
* Fix ONNX `RuntimeError` for OBB models exported with NMS by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22079
* Fix TensorRT dynamic shape calculation by Y-T-G in https://github.com/ultralytics/ultralytics/pull/22080
* Fix `wandb` logging twice when running DDP training by Laughing-q in https://github.com/ultralytics/ultralytics/pull/22073
* `ultralytics 8.3.200` Refactor Sony IMX exports to `imx.py` by Laughing-q in https://github.com/ultralytics/ultralytics/pull/22005

New Contributors
* miles-deans-ultralytics made their first contribution in https://github.com/ultralytics/ultralytics/pull/22063

**Full Changelog**: https://github.com/ultralytics/ultralytics/compare/v8.3.199...v8.3.200

8.3.199

📊 Key Changes
- Lazy model loading for faster imports (primary)
- Models like YOLO, SAM, RTDETR, NAS, YOLOE, FastSAM, YOLOWorld are now loaded on first access via `__getattr__`, preserving the public API. About ~3% faster `import ultralytics`.  
 See PR: [3% Faster Imports with Lazy Loading (21985)](https://github.com/ultralytics/ultralytics/pull/21985) by RizwanMunawar
- More consistent export outputs
- Quantized export NMS wrapper now returns unpackable tensors `(boxes, scores, labels, n_valid)` for non-keypoint tasks; keypoint outputs unchanged.  
 See PR: [Fix imx object detection export outputs (22045)](https://github.com/ultralytics/ultralytics/pull/22045) by Laughing-q
- Smarter TensorRT installation on Linux
- Auto-installs the CUDA-matching TensorRT wheel (e.g., `tensorrt-cu12`) and blocks known-bad versions for more reliable exports.  
 See PR: [Specify CUDA version during TensorRT installation (22060)](https://github.com/ultralytics/ultralytics/pull/22060) by Y-T-G
- Safer torch.compile defaults
- `attempt_compile()` now warns on `mode="max-autotune"` and uses `max-autotune-no-cudagraphs` instead; docs updated accordingly.  
 See PR: [Add warning and default to no-cudagraphs (22040)](https://github.com/ultralytics/ultralytics/pull/22040) by Y-T-G
- Clearer hyperparameter tuning plots
- New default filters out zero-fitness points in `plot_tune_results(..., exclude_zero_fitness_points=True)` for cleaner visuals.  
 See PR: [Exclude zero-fitness points in Tuner plots (22047)](https://github.com/ultralytics/ultralytics/pull/22047) by glenn-jocher
- Robustness fix for custom model parsing
- Prevents undefined `scale` errors in `parse_model()` when `scales` isn’t provided.  
 See PR: [Fix undefined variable in parse_model (22054)](https://github.com/ultralytics/ultralytics/pull/22054) by Y-T-G
- GPU test coverage re-enabled
- ONNX export with NMS for OBB re-enabled; CUDA export tests and GPU benchmarks now run when GPUs are available.  
 See PR: [Re-enable TensorRT export in GPU tests (22062)](https://github.com/ultralytics/ultralytics/pull/22062) by Laughing-q
- Docker docs modernized for NVIDIA Container Toolkit
- Replaces deprecated NVIDIA Docker approach with NVIDIA Container Toolkit; adds distro-specific install steps and standardizes `--runtime=nvidia`.  
 See PRs: [Docker Quickstart update (21994)](https://github.com/ultralytics/ultralytics/pull/21994), [Standardize GPU Docker commands (#22052)](https://github.com/ultralytics/ultralytics/pull/22052) by onuralpszr
- CI reliability and maintenance
- GPU runner updated to `gpu-latest`; Slack alerts now target specific failed jobs; runner image version parameterized.  
 See PRs: [Update GPU runner label (22051)](https://github.com/ultralytics/ultralytics/pull/22051) by glenn-jocher, [Refine Slack notifications (#22012)](https://github.com/ultralytics/ultralytics/pull/22012) by lakshanthad, [Parametrize runner version in Dockerfile (#22049)](https://github.com/ultralytics/ultralytics/pull/22049) by glenn-jocher
- New reference docs for lazy imports
- Adds a reference page explaining lazy imports in `ultralytics/__init__.py`.

🎯 Purpose & Impact
- Faster startup and the same API 🏎️
- Importing Ultralytics is quicker with zero code changes. You can still do:
 - `from ultralytics import YOLO`
 - `ultralytics.YOLO("yolo11n.pt")`
- More reliable deployment pipelines 🧰
- Standardized NMS export outputs simplify integration with ONNX/TensorRT and downstream code.
- Correct TensorRT package selection reduces install/export friction on Linux.
- Safer compilation defaults 🛡️
- `torch.compile` now prefers `max-autotune-no-cudagraphs`, avoiding CUDA Graphs issues while keeping performance benefits.
- Cleaner experiment insights 📈
- Tuning plots focus on meaningful runs by default, making it easier to spot what works.
- Improved docs and GPU usability 🧪
- NVIDIA Container Toolkit guidance and consistent `--runtime=nvidia` examples make GPU containers more predictable across distros.
- Better CI signal and stability 🔔
- Targeted Slack alerts and updated runners improve reliability without affecting user-facing features.

Helpful snippets:
- Import remains the same:
python
from ultralytics import YOLO
model = YOLO("yolo11n.pt")

- Tuner plots with zero-fitness points visible (previous behavior):
python
from ultralytics.utils.plotting import plot_tune_results
plot_tune_results("tune_results.csv", exclude_zero_fitness_points=False)

- GPU Docker run examples:
bash
sudo docker run -it --ipc=host --runtime=nvidia --gpus all ultralytics/ultralytics:latest


What's Changed
* docs: 📝 Update Docker Quickstart Guide to include NVIDIA Container Toolkit by onuralpszr in https://github.com/ultralytics/ultralytics/pull/21994
* Fix `imx` object detection export outputs by Laughing-q in https:

@pyup-bot pyup-bot mentioned this pull request Sep 29, 2025
Closed
@dagshub
Copy link

dagshub bot commented Sep 29, 2025

Join the discussion on DagsHub!

@pyup-bot
Copy link
Collaborator Author

pyup-bot commented Oct 6, 2025

Closing this in favor of #628

@pyup-bot pyup-bot closed this Oct 6, 2025
@guysmoilov guysmoilov deleted the pyup-scheduled-update-2025-09-29 branch October 6, 2025 14:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Reviewers

No reviews

Assignees

No one assigned

Labels

None yet

Projects

None yet

Milestone

No milestone

Development

Successfully merging this pull request may close these issues.

2 participants

@pyup-bot