TensorFlow MRI 0.16.0
Release 0.16.0
Breaking Changes
-
tfmri.convex
:- Several of the inputs and outputs of
admm_minimize
have been renamed
to improve clarity and to make the interface more consistent with the
tfmri.optimize
module.
- Several of the inputs and outputs of
Major Features and Improvements
-
tfmri.convex
:-
admm_minimize
now supports batches of inputs. -
admm_minimize
has two new argumentsf_prox_kwargs
and
g_prox_kwargs
that allow passing additional arguments to the proximal
operators off
andg
. -
admm_minimize
has a new argumentname
that allows specifying
the name of the operation. -
Method
_prox
ofConvexFunctionQuadratic
has a new argument
solver_kwargs
that allows passing additional arguments to the
internal solver. -
New properties
shape
andbatch_shape
forConvexFunction
and
its subclasses. These allow retrieval of static shape information. -
New methods
ndim_tensor
,shape_tensor
andbatch_shape_tensor
forConvexFunction
and its subclasses. These allow retrieval of the
dynamic shape information.
-
-
tfmri.linalg
:- New argument
bypass_gradient
forconjugate_gradient
.
- New argument
-
tfmri.recon
:lstsq
has a new argumentreturn_optimizer_state
which allows the
user to retrieve the internal optimizer state.
-
tfmri.optimize
:- New function
gradient_descent
implementing the gradient descent method
for minimization of differentiable functions.
- New function
-
tfmri.plot
:- New argument
dpi
for functionsimage_sequence
and
tiled_image_sequence
. - New function
close
, alias ofmatplotlib.pyplot.close
.
- New argument
Bug Fixes and Other Changes
-
tfmri.convex
:- Fixed a bug in method
prox
ofConvexFunctionNorm
,
ConvexFunctionL2Squared
andConvexFunctionQuadratic
that caused
errors when running in graph mode.
- Fixed a bug in method
-
tfmri.linalg
:- Fixed a bug in internal linear algebra framework that would cause errors
when running in graph mode.
- Fixed a bug in internal linear algebra framework that would cause errors
-
tfmri.plot
:- Removed lazy import mechanism which was causing problems when using
matplotlib
outside TFMRI. For now we use regular imports.
- Removed lazy import mechanism which was causing problems when using