Release MyGrad 1.9.0

v1.9.0

release 1.9.0

Release MyGrad 1.8.1

See changelog

Release MyGrad 1.8.0

add ver 1.8 docs (#264)

* add ver 1.8 docs

* Update docs/source/random.rst

Co-authored-by: Darshan Krishnaswamy <darsh797@gmail.com>

* move test functions; get rid of deprecated random_integers()

Co-authored-by: Darshan Krishnaswamy <darsh797@gmail.com>

Release MyGrad 1.7.1

Fixes a bug in :func:~mygrad.nnet.losses.negative_log_likelihood, where setting constant=True had no effect.

Release MyGrad 1.7.0

This release continues the process of integrating functions from mynn <https://github.com/davidmascharka/MyNN>_.

New features:

• Adds :func:~mygrad.nnet.initializers.glorot_normal
• Adds :func:~mygrad.nnet.initializers.glorot_uniform
• Adds :func:~mygrad.nnet.initializers.he_normal
• Adds :func:~mygrad.nnet.initializers.he_uniform
• Adds :func:~mygrad.nnet.initializers.normal
• Adds :func:~mygrad.nnet.initializers.uniform
• Adds :func:~mygrad.nnet.losses.focal_loss
• Adds :func:~mygrad.nnet.losses.negative_log_likelihood

Big thanks to David Mascharka!

Improvements:

The interfaces to :func:~mygrad.reshape and :func:~mygrad.Tensor.reshape were adjusted to match exactly the interfaces to their NumPy counterparts.
I.e. :func:~mygrad.reshape now requires newshape to be a sequence, whereas :func:~mygrad.Tensor.reshape can accept an unpacked sequence for its
newshape.

:func:~mygrad.Tensor.shape is now settable - triggering an in-place reshape of a tensor, matching the corresponding behavior in NumPy.

Internal changes:

The logic for writing an in-place operation has been consolidated into a convenient wrapper: :func:~mygrad.Tensor._in_place_op.

Release MyGrad 1.6.0

v1.6.0

release mygrad 1.6