ROCm/jax

Pull Requests Commits

Bump jaxlib version in README to 0.1.45

skye committed 5 years ago

6db1f0c0

try optimize=True with einsum

mattjj committed 5 years ago

783e24fd

attempt to fix changelog formatting bugs

mattjj committed 5 years ago

18f96742

autodiff cookbook: assume continuous second derivatives

mattjj committed 5 years ago

Verified 964cf4fb

Bump jaxlib version to 0.1.45 and update WORKSPACE and CHANGELOG. (#2785)

skye committed 5 years ago

Verified 4703b0fa

Merge pull request #2752 from apaszke/ad-fixme

mattjj committed 5 years ago

Verified 0d7d6637

Merge branch 'master' into ad-fixme

mattjj committed 5 years ago

Verified 18b23404

Merge pull request #2763 from jacobjinkelly/adamax

mattjj committed 5 years ago

Verified f527ed41

update version and changelog for pypi

mattjj committed 5 years ago

4433ddfb

also skip jax.numpy.unique test on tpu

mattjj committed 5 years ago

94178d98

Skip jnp.unique test on GPU (#2780)

jekbradbury committed 5 years ago

Verified 65dd1ef8

Fix distribution name in docstring (#2764)

grisaitis committed 5 years ago

Verified 2bc3c798

Add pmean to lax documentation (#2778)

jekbradbury committed 5 years ago

Verified 3f597833

Implement np.unique (#2760)

oliverastrand committed 5 years ago

Verified b36d1bff

Fix confusing documentation typo. (#2773)

lucasb-eyer committed 5 years ago

Verified 685c0de9

Delete cotangent references on their last use (#2719)

apaszke committed 5 years ago

Verified 88c802af

Added FAQ entry about relationship between VJP and JVP (#2762)

gnecula committed 5 years ago

Verified ff24f9a2

add adamax test

jacobjinkelly committed 5 years ago

61fc2bf2

jacobjinkelly committed 5 years ago

80789738

Fixes in the FAQ for RST (#2761)

gnecula committed 5 years ago

Verified 3ca7f6e4

Update automatic jaxlib install command to use nvidia-smi instead of nvcc. (#2758)

skye committed 5 years ago

Verified a1b4fe47

skip two unreliable tests

froystig committed 5 years ago

30c96ac3

Merge pull request #2684 from hexahedria/index_sugar

mattjj committed 5 years ago

Verified 5ab00e75

Raise an error if stop_gradient is called on non-arrays (#2750)

shoyer committed 5 years ago

Verified e6f0b8d8

Assert that reduction computations don't have constants. (#2754)

hawkinsp committed 5 years ago

Verified 9a5b8d62

Add a simple form of partial evaluation for while_loop. (#2497)

gnecula committed 5 years ago

Verified 7d716b83

Add FIXMEs for AD type errors

apaszke committed 5 years ago

8c00b35f

set precision=HIGHEST only for TPU test

jakevdp committed 5 years ago

aeb0d036

Thread precision through np.convolve & np.correlate

jakevdp committed 5 years ago

3265b8c3

danieldjohnson committed 5 years ago

835949a3

Newer Older