TPU errors?? Anyone else correct this yet?

[jimb2834]

Hello,

I was trying to test and got this error. I am using Colab Pro+ and have a TPU/High Mem.

Any idea why this occurs?

2022-12-22 13:40:59.434952: E external/org_tensorflow/tensorflow/compiler/xla/stream_executor/cuda/cuda_driver.cc:267] failed call to cuInit: CUDA_ERROR_NO_DEVICE: no CUDA-capable device is detected
WARNING:jax._src.lib.xla_bridge:No GPU/TPU found, falling back to CPU. (Set TF_CPP_MIN_LOG_LEVEL=0 and rerun for more info.)

[mar-muel]

Hi - unfortunately the code was only tested on Cloud TPU VMs which behave differently from Colab TPUs. At the very least you need to initialize the TPUs first on Colab, but I'm not sure this would be enough to make training work. Please share your solution in case it ends up working!

[jimb2834]

Hello, Yes did try that already on Coab, added TPU and HIghmem. Will have to get a TPU VM then and give it a try, Thanks for the reply

…

On Thu, Dec 22, 2022 at 5:48 AM Martin Müller ***@***.***> wrote: Hi - unfortunately the code was only tested on Cloud TPU VMs which behave differently from Colab TPUs. At the very least you need to initialize the TPUs first on Colab, but I'm not sure this would be enough to make training work. Please share your solution in case it ends up working! — Reply to this email directly, view it on GitHub <#8 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AMYPQADSQI7S5XP7DGQU2JLWORL2NANCNFSM6AAAAAATGXVAAA> . You are receiving this because you authored the thread.Message ID: ***@***.***>

-- *Jim Barnes,*