Y
Hacker News
new
|
ask
|
show
|
jobs
by
argee
12 days ago
That’s the problem. If all (or zero) models were trained on it, it would be fine as a benchmark.