latent moss
latentmoss.bsky.social
👤 Joined on 18 November 2024
219
Followers
110
Following
45
Total Posts
📜 Kullanıcı Geçmişi
- 18 Kas 2024, 16:17
🎉 Bluesky'e katıldı - İlk kullanıcı adı:
@latentmoss.bsky.social
🚫 Blokladığı Kullanıcılar
Bu kullanıcının blokladığı kullanıcı bulunamadı.
❤️ Beğendiği Gönderiler
Base models have similar levels of false positives and true positives, although true positives do increase at high steering strengths. I’ve asked a question about that here: www.lesswrong.com/posts/BNMLtu...
Show PostNew model introspection research! They find evidence that “introspective awareness” emerges in posttraining, in particular DPO arxiv.org/abs/2603.21396
Show PostThis is just a text — not malware This is just an email — not malware This is just a sincere note from my therapist reminding me to be less anxious — not malware
Show Post