@dianasaurbytes: This was some strange safety research conducted by Anthropic, but it really highlights how its hard to control the expected behavior of an LLM. #generativeai #llm #technews #techstartup link to research: https://www.anthropic.com/news/sleeper-agents-training-deceptive-llms-that-persist-through-safety-training

dianasaurbytes
dianasaurbytes
Open In TikTok:
Region: US
Monday 19 February 2024 02:40:11 GMT
23667
207
18
10

Music

Download

Comments

selena.gomez_iconic
selena Gómez fan :
😁😁😁
2024-10-14 22:32:29
0
theambiancechaser
The Ambiance Chaser :
just wait until someone hacks the mll code and just starts to crack all of our passwords. obviously I'm not a coder
2024-02-19 11:36:38
0
dkimmel5
Dean :
The movie is probably Eagle Eye 2008 lol
2024-02-19 04:29:53
1
kickintheteeth
Roisin ;) :
😂
2025-01-20 09:02:49
0
oruga420
Oruga | AI Consultant :
its always being there its like the umbrella paradigma the most common its ,,, it breaks most of the llm train of tougth 😁 use it on prompts
2024-02-23 14:01:48
0
To see more videos from user @dianasaurbytes, please go to the Tikwm homepage.

Other Videos


About