PyTorch, TensorFlow, JAX and NumPy — all of them natively using the same code
MIT License
Bot releases are hidden (Show)
float64()
conversion (thanks @eserie)slogdet()
implementation (thanks @eserie)__matmul__
(thanks @eserie)topk()
implementation (thanks @mglisse)Published by jonasrauber about 4 years ago
Published by jonasrauber about 4 years ago
prod()
Published by jonasrauber over 4 years ago
squeeze
now raises an error if a concrete axis is specified and this axis does not have length 1
Published by jonasrauber over 4 years ago
sort
pow
now accepts tensors for the exponentPublished by jonasrauber over 4 years ago
specified version numbers for dependencies
Published by jonasrauber over 4 years ago
Fixed PyPI compatibility
Published by jonasrauber over 4 years ago
flatten()
with additional start
and end
parameterstensor[boolean_mask]
Published by jonasrauber over 4 years ago
Tensor.__iter__()
is now implemented (instead of relying on __getitem__
for iteration, which was inconsistent)norms.l0
now returns the same dtype as the input (previously returned int, being inconsistent with other lp norms)mean
now requires float inputs for all frameworks (previously only for PyTorch and TensorFlow) and throws an appropriate error rather than crashingsum
is now guaranteed to work with int and float tensors as wellsum
for boolean TensorFlowTensor
instancesPublished by jonasrauber over 4 years ago
Tensor.numpy()
now returns a read-only array for all frameworks (previously this was inconsistent between frameworks and caused bugs because the numpy array may share memory with the actual tensor)Published by jonasrauber over 4 years ago
int
(in particular -1
), not just tuplesPublished by jonasrauber over 4 years ago
Tensor.__getitem__
now supports NumPy arrays (ndarray
)Published by jonasrauber over 4 years ago
TensorFlowTensor.__getitem__
with slices
Published by jonasrauber over 4 years ago
Published by jonasrauber over 4 years ago
Tensor.norms.*
methods
x.norms.l2()
where x
is an ep.Tensor
instancePublished by jonasrauber over 4 years ago
*args, **kwargs
)Published by jonasrauber over 4 years ago
__getitem__
now supports range
and list
take_along_axis
(for the last axis only)Published by jonasrauber over 4 years ago
ep.Tensor
now points to the base classep.TensorType
now points a TypeVar
bounded by ep.Tensor
astensors
, astensor_
and astensors_
restore_type
function that inverts the astensor
operationPublished by jonasrauber over 4 years ago
Tensor.tensor
is now called Tensor.raw
Published by jonasrauber over 4 years ago
TensorFlowTensor
now supports automatic casting to the common dtype
for all comparison operators