jakevdp
Follow
jakevdp
Follow
Achievements
Achievements
Highlights
- 3 discussions answered
Pinned
1,317 contributions in the last year
Less
More
Contribution activity
August 2021
Created 45 commits in 2 repositories
Created 2 repositories
- jakevdp/probability Jupyter Notebook
- jakevdp/.github
Created a pull request in google/jax that received 5 comments
Add jax.debugging.make_hlo()
Based on code by @levskaya See the generated docs here: https://jax--7719.org.readthedocs.build/en/7719/_autosummary/jax.debugging.make_hlo.html#ja…
+119
−0
•
5
comments
Opened 47 other pull requests in 2 repositories
google/jax
4
open
42
merged
- nansum/nanprod: add missing dtype validation
- check static arguments in numpy nullary functions
- Require array-like inputs for jnp.poly functions
- CI: remove flake8 from test requirements.
- jnp.histogramdd: fix support for range argument
- maml.ipynb: fix sync issue
- import tree_util in the JAX namespace
- re-enable test for jnp.linalg.lstsq
- CI: bump mypy & flake8 versions to newest
- Specify scipy in setup.py install_requires
- cleanup: remove redundant entry from test-requirements
- cleanup: remove some unnecessary test boilerplate
- Specify weak_type in DeviceArray repr
- DOC: fix code formatting in type_promotion.rst
- Require array-like inputs for jax.numpy set-like operations
- Cleanup: remove unused import
- logsumexp: fix issue with debug_nans
- Adjust wording for errors involving array rank
- Fix logsumexp issue with debug_nans and disable_jit
- scipy.stats.beta: fix pdf for x=0, 1
- jax.numpy: add missing x.choose() method
- mypy export of jax.numpy namespace
- Add optional fill_value argument to jnp.nonzero
- jax.image: fix error with zero-size image resizing
- [sparse] BCOO: convert inputs to jnp.array when possible
- Some pull requests not shown.
tensorflow/probability
1
closed
Reviewed 19 pull requests in 1 repository
google/jax
19 pull requests
- Check for aval weak_type attr before accessing it.
- Add np.polydiv numpy function
- jnp.histogramdd: fix support for range argument
- jnp.where: provide fill_value for single-argument form
- add support for scipy.special.{expn,expi,exp1}
- Specify weak_type in DeviceArray repr
- Bump styfle/cancel-workflow-action from 0.8.0 to 0.9.1
- Bump flatbuffers from 1.12 to 2.0
- Bump pre-commit/action from 2.0.0 to 2.0.3
- Add documentation for jax.device_get
- Complex-valued variance scaling initializers
-
Default to
jnp.float_type innn.initializers - Support scipy.fft.dct/dctn type=2
- Store avals in literals to avoid recanonicalization
- Do not unflatten trees with None values in grad.
- [sparse] add sparsify rule for lax.cond
- bugfix for sm3 optimizer
- DOC: many small fixes
- Perform float16 dot()s in float32 on CPU.
Created an issue in google/jax that received 2 comments
Regularize jax.numpy API
I've been picking away at this for the past year or so (since #3038), but I wanted to track it a bit more formally here. We want the jax.numpy API …
2
comments