4 ms·
IMHO the better middle ground is to use a nice (potentially fine tuned) small model locally, of which there are many now thanks to Chinese AI firms.
by yunohn 9mo ago
IMHO the better middle ground is to use a nice (potentially fine tuned) small model locally, of which there are many now thanks to Chinese AI firms.
- ithkuil 9mo agoAn expensive model can generate the training dataset