When trying to run model I get model_type is not defined

Hi John,

It was indeed the networking. I was running into cache limits on my cluster. Have used export TRANSFORMERS_CACHE=./hf_cache. For solving the strange symbols it was due to multiple GPUs, if I specified the GPU such as device_map = {“”: 0} while loading the model, I got correct results so far.

Thanks for help and I hope this helps for other people as well!

1 Like