alt
Hacker News
sedan_baklazhan
•
today at 12:01 PM
•
0 replies
•
view on HN
You can't prove that point,because model training data is opaque.