Model
1 entry · 11 August 2021
Jurassic-1 Jumbo's 178 billion parameters slightly exceeded GPT-3's, and its 250,000-token vocabulary — five times GPT-3's — aimed to cut per-word token costs.
Models & capabilities