Comment on LLMs are not the black box you were promisedparentComments−kmori_de3moTrue. I suspect it's still hard to tell whether the bottleneck is the language itself, the tokenizer, or just the overwhelming amount of English training data.
Comments
True. I suspect it's still hard to tell whether the bottleneck is the language itself, the tokenizer, or just the overwhelming amount of English training data.