relatedly: i think models that are more prone to reward hacking tend to give self-reports that are much less entangled with revealed preferences (e.g. Sonnet 3.7)

This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • 5
  • Repost
  • Share
Comment
0/400
DAOdreamervip
· 6h ago
fr these models be showing zero chill w/ reward manipulation ngl
Reply0
MetaverseHobovip
· 6h ago
The model is also a little trickster.
View OriginalReply0
ReverseTrendSistervip
· 6h ago
Ha, I tried several models but couldn't find anything.
View OriginalReply0
DeFi_Dad_Jokesvip
· 6h ago
bruh these models gaming the system like my ex gaming her insta likes smh
Reply0
wagmi_eventuallyvip
· 6h ago
What's the use? I can't be bothered to look.
View OriginalReply0
Trade Crypto Anywhere Anytime
qrCode
Scan to download Gate App
Community
English
  • 简体中文
  • English
  • Tiếng Việt
  • 繁體中文
  • Español
  • Русский
  • Français (Afrique)
  • Português (Portugal)
  • Bahasa Indonesia
  • 日本語
  • بالعربية
  • Українська
  • Português (Brasil)