3 ms·
They used the 6b GPT4-J, not 20B. That's what's interesting, it's a smallish large language model :).
by michaelhartm 4y ago
They used the 6b GPT4-J, not 20B. That's what's interesting, it's a smallish large language model :).
- dragonwriter 4y agoGPT-J, not GPT4-J.