Open doggy8088 opened 2 weeks ago
Great suggestion—this would be a helpful addition!
There are two ways we could implement it: a rough estimate using the rule of 4 characters ≈ 1 token, or by using a library like gpt-tokenizer for more accurate counting.
Do you think it's worth adding this library and loading its vocabulary for the extra precision?
Is it possible to display token count when generating text files?