array-api-compat icon indicating copy to clipboard operation
array-api-compat copied to clipboard

Compatibility layer for common array libraries to support the Array API

Results 59 array-api-compat issues
Sort by recently updated
recently updated
newest added

See https://github.com/scipy/scipy/issues/21054#issuecomment-2192403563

There seem to be some issues with the signatures of functions wrapped by `get_xp`. I haven't narrowed down the exact problem, but here's an MRE: ```python3 import cupy as xp...

Medium Priority

IIUC the standard states that input dtype is to be preserved for floating point types when `dtype` is unspecified. > **dtype** - data type of the returned array. If `None`,...

Follow-up to https://github.com/data-apis/array-api-compat/issues/144#issuecomment-2161491665 Sometimes to make existing code compatible with backends that are not fully standard compliant, we would need to create copies where the original code would not. For...

The array API standard seems to support negative step. > The basic slice syntax is `i:j:k` where `i` is the starting index, `j` is the stopping index, and `k` is...

Medium Priority
blocked by upstream

Bumps the actions group with 1 update in the / directory: [pypa/gh-action-pypi-publish](https://github.com/pypa/gh-action-pypi-publish). Updates `pypa/gh-action-pypi-publish` from 1.10.1 to 1.10.3 Release notes Sourced from pypa/gh-action-pypi-publish's releases. v1.10.3 💅 Cosmetic Output Improvements In...

``` /opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/array_api_compat/common/_helpers.py:516: in array_namespace from .. import torch as torch_namespace /opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/array_api_compat/torch/__init__.py:15: in from ._aliases import * # noqa: F403 /opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/array_api_compat/torch/_aliases.py:14: in from ._info import __array_namespace_info__ _ _ _ _ _...

Since the 2022.12 standard, the required implementation of `sign` has been: but I think only the most recent versions of libraries follow this (if any). Older versions of all libraries...

Follow-up to #288 Get mypy _mostly_ green everywhere except torch.

- On PyTorch >=2.3, support uint16, uint32, and uint64 in `result_type` - On PyTorch >=2.3, `sum` and `prod` will now promote uint8, uint16 and uint32 to uint64.

blocked by upstream