Hello All,
I am using a Mac with M1 processor and I am using a GPT-Neo model for my project like this:
from happytransformer import HappyGeneration
happy_gen = HappyGeneration(model_type="GPT-NEO", model_name="EleutherAI/gpt-neo-125M")
happy_gen.train("somefile.txt")
Is it possible to train HappyGeneration on M1 GPU so that it could be little bit faster? I am unable to find any option that specifies a device to be selected for training.