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
-
Google
- Seattle WA
- http://www.vanderplas.com
Pinned
1,516 contributions in the last year
Contribution activity
May 2020
Created a pull request in google/jax that received 4 comments
- Fix arr.view() on TPU
- Expand type support for random uniform() & randint()
- Add support for 8- and 16-bit output in _random_bits
- disable arr.view() test on TPU
- Add lax implementation of np.isin() and np.in1d()
- promote integer inputs to float in jnp.median() and jnp.quantile()
- Implement np.histogram_bin_edges and np.histogram
- Implement .view() method of jax.numpy arrays
- fix astype() test
- Update random.logistic() to prevent infinities
- Add lax implementation of np.trapz
- Modify linspace so that endpoints equal the inputs.
- Implement np.digitize
- Add Colab test notebooks for CPU, GPU, and TPU
- DOC: add unravel_index to docs/jax.numpy.rst (forgotten in #2966)
- Cleanup: move _wraps into jax.numpy._utils.
- Add implementation of np.searchsorted
- Deprecate random.shuffle() and implement random.permutation() for multi-dim inputs
- BFGS algorithm
- Remove workaround for Mac linear algebra bug that is fixed in the min…
- Add lax implementation of np.indices
- Clean up jax.ops namespace.
- Add jnp.unravel_index
- DOC: add a table of contents for top level API docs
- DOC: write a new dosctring for jax.numpy.vectorize
- Numpy union1d
- reduce use of lax on static data (e.g. shapes)
Created an issue in google/jax that received 4 comments
jax.numpy.linspace() does not faithfully recover endpoints
Example: >>> import numpy as np >>> import jax.numpy as jnp >>> a, b = -2.3328583, 2.8459014 >>> np.linspace(a, b, 10)[-1] == b True >>> jnp.linspace(
4
comments