keccers pfp
keccers
@keccers.eth
“We argue that a lightweight, five-step Cognitive-Behavioural Therapy (CBT) loop—inserted inside or immediately above every system prompt—offers a practical defence [against ai model hallucinations]” https://zenodo.org/records/15556365
6 replies
3 recasts
17 reactions

Eric Olszewski pfp
Eric Olszewski
@ericolsz
Some people wake up early to workout, others wake up early to raid ncbi's latest additions without posting a link in the comments 🔪
1 reply
0 recast
2 reactions

azb pfp
azb
@azbest
as we all should
1 reply
0 recast
2 reactions

Preator pfp
Preator
@fourvork
I still think it will eventually start hallucinating once you use too much tokens, but that's a great model
0 reply
0 recast
1 reaction

JAstra pfp
JAstra
@jastra
Even AI needs behavioural training
0 reply
0 recast
1 reaction

Ciberdelia pfp
Ciberdelia
@ciberdelia
Wow, thi is really very interesting
0 reply
0 recast
1 reaction