Y
Hacker News
new
|
ask
|
show
|
jobs
by
Majromax
155 days ago
If you remove the conditional and keep the same math, you divide by zero and get nans.
In the limit
as temperature goes to zero, you do in fact get maximum likelihood sampling.
1 comments
dnautics
155 days ago
if (t==0) argmax(logits) else pick(logits)
link
eru
155 days ago
You are ignoring the limit taking.
link
dnautics
154 days ago
If t < 0.1 return {error, 400}
link
eru
154 days ago
Do you know that a mathematical limit is?
link