Open aliaclarkelsayed opened 1 month ago
Maybe call the commands separately? Are you sure you installed the env?
I got the same error raised when running predict.py
from the script. Not sure if it's from the compatibility problem of jax
and flax
https://github.com/google/flax/issues/3180
Could you @patrickbryant1 specify your final working version of jax
and flax
? I would like to try using the same version. I'm using Cuda 12.2 btw if relevant
But for sure, pip install --upgrade numpy
in install_dependencies.sh
gave me error since numpy 2
just released.
Hi,
Maybe see some of the other errors/look in the env for the versions. FLAX is not used so idk if this is causing your issues?
I see, we will check the new versions and provide a new env in a couple of weeks.
Best,
Patrick
Hi, good to know flax is not used. Because it is listed in the environment.yml?
After running the following as in the test example:
conda activate umol bash predict.sh
I got the following error from configurations.py: AttributeError: 'Config' object has no attribute 'define_bool_state'