Closed tashrifbillah closed 2 years ago
Use of an advanced numpy only helps us bypass this warning and probably gives loss magnitude in better precision:
min3-tf-gpu/envs/gpu/lib/python3.6/site-packages/tensorflow/python/framework/dtypes.py:532: FutureWarning: Passing (type, 1) or '1type' as a synonym of type is deprecated; in a future version of numpy, it will be understood as (type, (1,)) / '(1,)type'. np_resource = np.dtype([("resource", np.ubyte, 1)])
All the memory leak error we got in the cluster so far should be due to the following numpy version:
https://github.com/pnlbwh/CNN-Diffusion-MRIBrain-Segmentation/blob/03108b1f49f2e2df62d6034d746a3626ccee9fa5/environment_gpu.yml#L13
Tashrif's independent CNN design found the above.
More investigations to follow ...