4 ms·
> use small LLM for quick response with something generic while big LLM computes Can't wait for poorly implemented chat apps to always start a response with "T
by rjtavares 3y ago
> use small LLM for quick response with something generic while big LLM computes
Can't wait for poorly implemented chat apps to always start a response with "That's a great question!"