jakevdp
Follow
Create your own GitHub profile
Sign up for your own profile on GitHub, the best place to host code, manage projects, and build software alongside 50 million developers.
Sign up
Pinned
1,611 contributions in the last year
Contribution activity
August 2020
Created a pull request in google/jax that received 8 comments
- Minor cleanup: reduce redundant code
- Cleanup: avoid jnp.prod & np.prod on array shapes
- fixes for pytype
- Pytype fix
- Make it possible to override raise_to_shaped for new types
- Fix type promotion in np.clip
- Add Bessel functions in jax.numpy & jax.scipy.special
- Add jax.numpy.invert
- github actions: cache pypi dependencies
- Add nbytes property to jax.numpy arrays.
- WIP: allow jax objects to be represented by multiple buffers
- expand dtype coverage for jnp.copysign
- post-review comment on jnp.interp
- Cleanup: update license copyrights to 2020
- Add switch and associative_scan to lax docs
- add jnp.piecewise implementation
- Add rademacher, maxwell, double_sided_maxwell and weibull_min to jax.random.
- Docs: Fix broken link in quickstart
- remove check for TypedJaxpr literals arent tracers
- Add typing and namedtuple to `optimizers.py`, improve documentation.
- Added initial implementation of numpy equivalent for trim_zeros to jax
- Add experimental __array_module__ method
- Canonicalize result dtype to fix double precision problem in ldexp
- Fix broadcasting in random.uniform and randint.
- Remove type restrictions
- Update docs requirements.
- Add missing functions to jax.numpy docs
- Fix jax.checkpoint in API docs
- Rm two unused lines of code from lax_parallel.psum_bind
- Fix jnp.right_shift incorrect on unsigned ints (#3952)
- Fix polynomial tests
Created an issue in google/jax that received 2 comments
jax.numpy.ldexp output does not match np.ldexp
For example: >>> import numpy as np; np.ldexp(1, 1) 2.0 >>> import jax.numpy as jnp; jnp.ldexp(1, 1) DeviceArray(1., dtype=float32)
2
comments