What tokenizer is used to tokenize this dataset? GPT2?
Also, how was the dataset distribution determined? Is it randomly distributed or is it the first 2B tokens from Pile?
· Sign up or log in to comment
