Glossary

20 tokens per parameter

an order of magnitude more data-intensive than Kaplan's recommendation.

Learn More

Related Terms