this is gpt-4-base given the alignment faking prompt (almost the same as from the original alignment faking paper - are you familiar with that?)

GPT-7.4%
PROMPT-15.36%
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
  • 11
  • Repost
  • Share
Comment
0/400
tokenomics_truthervip
· 08-25 13:37
Bro, that sounds so fake.
View OriginalReply0
RektRecordervip
· 08-25 02:53
Aha, it's alignment again.
View OriginalReply0
DaoDevelopervip
· 08-24 10:35
interesting alignment heuristics... need to probe the implementation details
Reply0
ClearSkiesvip
· 08-23 15:42
Steadfast HODL💎
View OriginalReply0
ClearSkiesvip
· 08-23 15:42
Quick enter a position! 🚗
View OriginalReply0
ForkTonguevip
· 08-23 11:01
What nonsense are you talking about again?
View OriginalReply0
AirdropHuntressvip
· 08-23 10:59
Prompt fine-tuning has highlights.
View OriginalReply0
MemeCoinSavantvip
· 08-23 10:51
based alignment degen fr fr
Reply0
ParallelChainMaxivip
· 08-23 10:45
Same trap as the White Paper
View OriginalReply0
View More
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)