Y
Hacker News
new
|
ask
|
show
|
jobs
by
meepmorp
997 days ago
I've seen 1.5 chars/token used as a rule of thumb for estimating token counts in Chinese text.