diff options
author | Benjamin Klimczak <benjamin.klimczak@arm.com> | 2022-11-28 13:36:03 +0000 |
---|---|---|
committer | Benjamin Klimczak <benjamin.klimczak@arm.com> | 2022-12-06 17:05:37 +0000 |
commit | 0241f96fce1bbfa11f21bfbfa161b51f3170a51b (patch) | |
tree | f6ceae63670dd58f0b58de0ea55e039eddca3b50 /setup.cfg | |
parent | 6a88ee5315b4ce5b023370c1e55e48bf9f2b6f67 (diff) | |
download | mlia-0241f96fce1bbfa11f21bfbfa161b51f3170a51b.tar.gz |
MLIA-669 Upgrade dependencies with Vela 3.6
With Vela 3.6 we are able to remove the special treatment of aarch64
in our dependencies, i.e.
- upgrade Vela to version 3.6 that resolves a compatibility issue for
aarch64 in 3.4 and 3.5.
- upgrade to TensorFlow 2.10 which now supports aarch64 (therefore
making it obsolete to use 'tensorflow-aarch64').
Change-Id: I86508b667b5ccb55bfd11dcae9defc54e5ef74de
Diffstat (limited to 'setup.cfg')
-rw-r--r-- | setup.cfg | 11 |
1 files changed, 2 insertions, 9 deletions
@@ -28,17 +28,10 @@ python_requires = >=3.8 package_dir = = src packages = find_namespace: -# On aarch64 machine, tensorflow does not pin the numpy version but -# vela does. The pre-built aarch64 tensorflow 2.8.2 uses a numpy whose -# version is conflicting with ethos-u-vela version 3.4.0. Therefore, -# we downgraded both tensorflow and vela versions to make sure there is -# no numpy version conflict. install_requires = - tensorflow-aarch64~=2.7.3; platform_machine=="aarch64" - tensorflow~=2.9.2; platform_machine=="x86_64" + tensorflow~=2.10.1 tensorflow-model-optimization~=0.7.3 - ethos-u-vela~=3.3.0; platform_machine=="aarch64" - ethos-u-vela~=3.5.0; platform_machine=="x86_64" + ethos-u-vela~=3.6.0 requests rich sh |