48 ms·
Actually, Mistral by virtue of not aligning/safety lobotomizing their models will indeed trivially "help in producing deadly gases or uranium enrichment facilit
by Der_Einzige 2y ago
Actually, Mistral by virtue of not aligning/safety lobotomizing their models will indeed trivially "help in producing deadly gases or uranium enrichment facilities!"
- Beretta_Vexee 2y agoIt's quite a feat to demonstrate that you know nothing about LLMs, chemistry or the nuclear industry in so few characters.
- Der_Einzige 2y agoLet's see your google scholar or huggingface profile. You better square up if you're going to talk shit. It's telling that the release from OpenAI today warns about exactly this threat in their lengthy security section: https://cdn.openai.com/o3-mini-system-card.pdf https://cdn.openai.com/o3-mini-system-card.pdf