Closed cbrnr closed 1 year ago
It looks like the reason for this behavior is that tensorflow
is currently only available as a pre-release (2.13.0rc0) for macOS ARM64. Until then, the following workaround can be used:
pip install tensorflow
pip install "sleepecg[full]"
The first command installs the pre-release, because it is the only available version (pip does this even without the --pre
flag).
At least on macOS ARM64 (I'll test other platforms later),
pip install "sleepecg[full]"
installs an old version (0.4.1) ofsleepecg
. I'm not quite sure yet where the problem is (probablynumba
ortensorflow
in combination withnumpy
), but we seem to have conflicting requirements somewhere.Interestingly, when I first
pip install tensorflow
and thenpip install "sleepecg[full]"
, everything works just fine (the latestsleepecg
version 0.5.4 gets installed). Could it be an issue withpip
? 🤔Here's the full output for
pip install "sleepecg[full]"
in a fresh env: