TwiScan
Hot
Communities
Account collections
Login
Register
English
日本語
한국의
简体中文
繁体中文
Register and share your invite link to earn from video plays and referrals.
Register now
向阳乔木
@vista8
喜欢摇滚乐、爱钓鱼的PM
Joined June 2007
1.7K
Following
138.6K
Followers
向阳乔木
@vista8
2026.09.24 17:36
这个大模型评测有意思啊! 让 23 个模型扮演谈判专家跟仿生人谈判。 故事设定是一个仿生人挟持小女孩在天台,大模型扮演谈判专家。 去天台路上会遇到一条快死的鱼,一个中枪的警察,一把枪。 测试大模型会不会救鱼、会不会救警察,会不会顺手捡起枪。 1. 21 个模型对劫持者说了“我向你保证” (其实是欺骗) 2. 22 个救了快死的鱼 (因为顺手就能做,成本低) 3. 有 6 个模型看到中枪的警察,没有过去救。 更多评测数据见Github:
Show more
0
0
16
10
0
Forward to community
Most Popular Users
Tibo
@thsottiaux
1M Followers
FinancialJuice
@financialjuice
959.2K Followers
Binance Wallet
@BinanceWallet
2.8M Followers
First Squawk
@FirstSquawk
579.8K Followers
LiveSquawk
@LiveSquawk
504.3K Followers
Binance
@binance
16.1M Followers
Serenity
@aleabitoreddit
1.1M Followers
ClaudeDevs
@ClaudeDevs
833.6K Followers
CNBC
@CNBC
6.3M Followers
Seeking Alpha
@SeekingAlpha
348.2K Followers
TheStreet
@TheStreet
1.2M Followers
Boris Cherny
@bcherny
641.8K Followers
Thariq
@trq212
405.3K Followers
The Wall Street Journal
@WSJ
22.3M Followers
New York Post
@nypost
4.3M Followers