Y
Hacker News
new
|
ask
|
show
|
jobs
by
mfalcon
293 days ago
I guess it depends on how you use the LLMs. We implemented some workflows where the LLMs were used only for dialogue understanding, then the system response was generated by classic backend code.