Found a BUG and basic docs code fails to run on kaggle tpu

So just to try TPU, i’ve used the code by @muellerzr here on kaggle tpu vm, it gives me
the error BrokenProcessPool: A process in the process pool was terminated abruptly while the future was running or pending. link to traceback and code
Any help on this is appreciated.

BUG
Also accelerate uses os.environ variable TPU_NAME to check if its tpu or not and it won’t work here,and this leads to trying to run 8 gpu’s in a TPU environment :smiling_face_with_tear: