Installing tensorflow-metal downgraded the python-six dependency from 1.16.0 to 1.15.0 because the dependency specification of tensorflow-metal specififed six~=1.15.0
. This is a pinned version which is something libraries like tensorflow-metal should avoid at all cost; what if I want to use tensorflow together with another library that requires six>=1.16
? I can't imagine that there is anything in six 1.16 that would break tensorflow.
Issue report: tensorflow-metal unnecessarily pins dependencies
Add a Comment