Kategori: AI
-
The majority of hallucination checks begin following the initial writing of an answer by the LLM. We then hold someone up to a second standard of judgment, or we hand over dangerous cases to a person. That works but that is looking at the solution only as one-off. What if the LLM generates five different…
Læs hele historien hos analyticsvidhya.com → -
This is a research update from a replication of Serrano et al., 2026. I’m releasing these results now because I hope it inspires others to run similar / followup experiments.tl;dr:If a model wants to subvert evaluation safeguards, it may choose to take misaligned actions very rarely. Because evaluation is much shorter than deployment, this could…
Læs hele historien hos lesswrong.com → -
Cross-posted from my personal blogThe success of RL in LLMs has been puzzling me for a while. People have developed various information-theoretic style arguments by which they argue that RL is extremely informationally inefficient compared to pretraining, that it can only impart a tiny amount of bits, and that it can only bring out behaviours…
Læs hele historien hos lesswrong.com → -
FF EAI Robotics Ecosystem Inc. and Faraday Future Intelligent Electric Inc. announced on October 15, 2026 the joint launch of FF Nova, an entry-level embodied AI humanoid robot priced at $2,490, and FF Nova Mini, a $299 humanoid for foundational Physical AI education, alongside a new curriculum and developer programs. Both robots join the "One…
Læs hele historien hos unite.ai → -
Based on a recent paper, training models to give bad advice in a narrow domain doesn't make the model or "the persona" broadly evil in a robust way. It just makes the particular system prompt[1] tokens you fine-tuned with evoke an evil persona, and perturbing it a little (e.g. changing the date slightly) substantially reduces…
Læs hele historien hos lesswrong.com → -
Meta released Muse Spark Open, an 816B-parameter open-weight MoE model with 42B active parameters and a 1M-token context. It leads open models on MCP-Atlas and runs on 2 M5 Ultra Mac Studios. The post Meta Releases Muse Spark Open: 816B Open-Weight MoE With 1M Context appeared first on MarkTechPost.Læs hele historien hos marktechpost.com →
Læs hele historien hos marktechpost.com → -
NVIDIA, working with IIT Madras and its AI4Bharat lab, on October 16, 2026 introduced Nemotron-pretraining-Indic-v1, a synthetic pretraining corpus of approximately 2.8 trillion tokens across 22 Indian languages that is releasing soon, together with the open, GPU-accelerated data generation pipeline used to create it. The 22 languages are those listed in India's Constitution, which the…
Læs hele historien hos unite.ai → -
Work done as part of the MATS 9.1 extension program, mentored by Richard Ngo. 0: IntroductionA previous post of mine presents a toy model of self-sabotage in intrapersonal games. This post applies the idea of cross-temporal incoherence to multistage games with observable actions. These games are reinterpreted with new equilibria,[1] defined through variants of the…
Læs hele historien hos lesswrong.com → -
Læs hele historien hos huggingface.co →Læs hele historien hos huggingface.co →
Læs hele historien hos huggingface.co → -
This post was first drafted as part of the ML4Good bootcamp.Would it be possible to implement a corrigible agent? Is that likely to happen given current trajectories? Historically, the answer was a resounding "no", but with more recent theories about how LLMs learn, corrigibility now seems possible, if we're careful.DefinitionsFirst, let's define some terms.AgentAn agent…
Læs hele historien hos lesswrong.com →