a collection of custom builds of latest tensorflow with avx2 support
link | TF | PY | CPU/GPU |
---|---|---|---|
fo40225 | 1.4-1.9 | 36 | CPU, GPU |
aluo-x | 1.5 | 36 | CPU |
faisalthaheem | 1.4-1.8 | 35 | CPU |
link | TF | PY | CPU/GPU |
---|---|---|---|
lakshayg | 1.1-1.10 | 35,36 | CPU,GPU |
inoryy | 1.10-1.13,2.0 | 36,37 | CPU |
bukalapak | 1.10 | 35,36 | CPU |
-
To view which wheel your platform supports, you can run blew code.
Run
from pip._internal.pep425tags import get_supported as f; print(f())
If failed try
from pip.pep425tags import get_supported as f; print(f())
-
Do not rename whl filename e.g.
..._amd64.whl
to..._amd64(avx2).whl
, or rename back before installation. Thepip
identifies the file by its filename. Otherwise you will fail with "whl is not a supported wheel on this platform
".