Jake Vanderplas
Jake Vanderplas
### Describe the issue: Starting in numpy 1.25.0, it appears that the `ufunc.reduce` method does not recognize the `identity` argument for ufuncs created with `np.frompyfunc` ### Reproduce the code example:...
cc/ @mattjj, @froystig ~Test failures are due to the fact that many of our tests reuse keys!~ ~Test failures caused by key reuse are now addressed via explicit calls to...
The previous description is getting less relevant, because Autograd is arguably less well-known than JAX itself! cc/ @mattjj, as discussed earlier.
NumPy 1.22 was released in December 2021, so we are now past the limit of our [deprecation policy](https://jax.readthedocs.io/en/latest/deprecation.html).
It looks like only the number of nodes is validated, meaning you can transpose trees that have the wrong input structure: ```python import jax outer_treedef = jax.tree_util.tree_structure(['*', '*', '*']) inner_treedef...
With x64 mode disabled, type promotion for 32-bit unsigned ints can lead to data corruption: ```python from jax.numpy.lax_numpy import _promote_dtypes import numpy as np a = np.uint32(2 ** 32 -...
The `v2022-12` tag is broken due to the bug fixed in https://github.com/data-apis/array-api/pull/455, which as far as I can tell has not yet been made part of a release or tag....
Often when I'm developing a new package, I want a very specific jupyter kernel or set of kernels containing combinations of dependencies I only use for that project, but it's...
This is the result of an initial end-to-end readthrough of the new beginning tutorial content. Eventually this will replace the existing JAX 101 docs and the hodgepodge of "getting started"...
Example; when running this file: ```python # test.py import jax jax.config.update('jax_enable_key_reuse_checks', True) def f(seed): key = jax.random.key(seed) x = jax.random.uniform(key) y = 2 * x y += jax.random.normal(key) return 2...