Release v2.3.0

• Adds support for NumPy 2.0
• Minimum supported NumPy version is now 1.24
• Minimum supported Python version is now 3.9
• Adds testing for Python 3.12
• Adopts PyPI Trusted Publishers method for releasing new versions -- for improved security.

Release v2.2.0

What's Changed

• Improve nnet.initializers docs by @davidmascharka in #400
• remove unnecessary wraps from uber by @rsokl in #401
• Bump numpy from 1.18.1 to 1.21.0 in /docs by @dependabot in #402
• Fix tests by @rsokl in #403
• Bump numpy from 1.21.0 to 1.22.0 in /docs by @dependabot in #404
• Add Ellipsis to IndexType and permit arbitrary-length tuples in index by @rsokl in #406
• Drop support for numpy 1.19 by @rsokl in #405
• Simplify back-prop scheme by @rsokl in #407
• Touchup internal annotations by @rsokl in #408
• remove bad use of wraps in test by @rsokl in #409
• Improve test suite and CI by @rsokl in #411
• Bump actions/checkout from 1 to 3 by @dependabot in #413
• Bump actions/setup-python from 2 to 4 by @dependabot in #412
• Bump sphinx-rtd-theme from 0.5.2 to 1.1.1 in /docs by @dependabot in #416
• Bump sphinx from 3.5.4 to 5.3.0 in /docs by @dependabot in #414
• Bump numpydoc from 1.1.0 to 1.5.0 in /docs by @dependabot in #415
• prepare new release by @rsokl in #420

New Contributors

• @dependabot made their first contribution in #402

Full Changelog: v2.1.0...v2.2.0

Release MyGrad 2.1.0

See detailed changelog here.

What's Changed

• update docs dependencies, fix deprecated css, fix doc typo by @rsokl in #376
• Use faster dict-construction syntax by @rsokl in #377
• Update default behavior of derivative of linalg.norm and abs at x=0 by @rsokl in #379
• add nightly tests against pre-releases by @rsokl in #380
• Adds mygrad.save and mygrad.load by @rsokl in #383
• accommodates hypothesis #3065 by @rsokl in #385
• remove bad example from docs; improve type annotations by @rsokl in #386
• Implement atleast_kd by @rsokl in #387
• Update test_duplicating_graph.py by @rsokl in #388
• test should not pass None to array(..., copy=<>) by @rsokl in #390
• fix broadcasting hack by @rsokl in #391
• NEP 29: min numpy is now 1.19 by @rsokl in #394
• add Tensor.index method by @rsokl in #395
• Implement clip(..., out=) by @rsokl in #396
• Docs for 2.1.0 by @rsokl in #397

Full Changelog: v2.0.2...v2.1.0

Release MyGrad 2.0.2

See changelog

Release MyGrad 2.0.1

See the changelog

Release MyGrad 2.0.0

Official release of MyGrad 2.0! 🎉🎉🎉

See the changelog for all of the details!

Release MyGrad 2.0.0.dev4

Adds convenient overrides and mirrors to non-differentiable numpy functions and (non-complex) dtypes.

• allclose
• bincount
• can_cast
• copyto
• may_share_memory
• min_scalar_type
• result_type
• shares_memory
• shape

Release MyGrad 2.0.0.dev3

Implements Tensor.__array_function__.

Common numpy functions now have "drop-in" autodiff, when they operate on a tensor:

>>> x = mg.tensor([2., -1., 0.])
>>> numpy.einsum("i,i", x, x).backward()
>>> x.grad
array([ 4., -2.,  0.])

Add ufunc overrides

Implements __array_ufunc__ to override/dispatch NumPy ufuncs via mygrad.Tensor

Release MyGrad 2.0.0.dev1

A pre-release of MyGrad 2.0.0