resources
#1
by
ArthurFischel
- opened
What kind of resources did it take to train this model?
I trained it on nvidia A4000 gpu for a few hours.
Thanks for taking the time to respond. How much of the pile did you train on? If you had to give a rough token estimate?
I did not measure that for this project; but you might be interested to look at this: https://huggingface.co/naxautify/pythia-1.4b-deduped-8k