eagerpy

New Features

• float64() conversion (thanks @eserie)
• slogdet() implementation (thanks @eserie)
• support for __matmul__ (thanks @eserie)
• topk() implementation (thanks @mglisse)

Bug Fixes

• fixed JAX array recognition (thanks @zimmerrol)
• minor changes to support TensorFlow compilation (thanks @eserie)

New Features

• indexing with multiple tensors now supports more cases
• massively improved docs and guides

New Features

• added prod()
• added development documentation
• improved the reliability of the tests
• other improvements and bug fixes

squeeze now raises an error if a concrete axis is specified and this axis does not have length 1

New Features

• added sort
• pow now accepts tensors for the exponent
• docs are now automatically build on GitHub

specified version numbers for dependencies

Fixed PyPI compatibility

New Features

• added flatten() with additional start and end parameters
• added boolean tensor indexing, i.e. tensor[boolean_mask]
• other improvements and bug fixes

New Features

• 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 crashing
• sum is now guaranteed to work with int and float tensors as well
• fixed a bug in sum for boolean TensorFlowTensor instances

New Features

• 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)

New Features

• reshape now accepts a single int (in particular -1), not just tuples

New Features

• Tensor.__getitem__ now supports NumPy arrays (ndarray)

Bug fixes

• fixed a bug in TensorFlowTensor.__getitem__ with slices

New Features

• bug fixes for Python 3.7

New Features

• added Tensor.norms.* methods
  • e.g. x.norms.l2() where x is an ep.Tensor instance
• other improvements to code, docs, tests

New Features

• fully specified all parameters with descriptive names (no more *args, **kwargs)
• fully type-annotated all functions and all tests
• improved docs
• bugfixes

New Features

• documentation
• performance optimizations
• __getitem__ now supports range and list
• added take_along_axis (for the last axis only)
• better type annotations

New Features

• more type annotations
• ep.Tensor now points to the base class
• ep.TensorType now points a TypeVar bounded by ep.Tensor
• added astensors, astensor_ and astensors_
  • the underscored variants return a restore_type function that inverts the astensor operation
• more tests
• stricter mypy config

New Features

• Tensor.tensor is now called Tensor.raw
• added type hints to basically all methods and framework functions
• refactored large parts of the code
• many other bug fixes and improvements following from that

New Features

• TensorFlowTensor now supports automatic casting to the common dtype for all comparison operators
• other improvements and bug fixes