Tootfinder

Neue Farbe für Wände: Nachhaltig durch Pilze und Algen?
Eine Tinte aus Pilzen und Algen soll Hauswände vor Schadstoffen schützen, CO₂ speichern und Risse reparieren. Das Projekt will Gebäude nachhaltiger machen.

🇺🇦 #NowPlaying on KEXP's #VarietyMix
Alison Goldfrapp:
🎵 Reverberotic
#AlisonGoldfrapp
https://open.spotify.com/track/047E83iJCGmU3Sq3kabk7G

¿Por qué le llamamos "arañazo" a los "rasguños" si las arañas no arañan? 🤯
¿O alguien acaso ha visto alguna vez a una araña arañando?
¿Quién me lo puede explicar? 🤔

🔊 #NowPlaying on #BBCRadio3:
#NewMusicShow
- Unsuk Chin's Xi, plus world premieres by Jessie Marino, Sylvia Lim and Alexander Papp
Kate Molleson presents Unsuk's Xi alongside the world premieres of Jessie Marino's No Salt, Sylvia Lim's Moss that Holds and Alexander Papp's deformed figures.
Relisten now 👇
https://www.bbc.co.uk/programmes/m002d1q2

The Cause of Alzheimer's Might Be Coming From Within Your Mouth : ScienceAlert
https://www.sciencealert.com/the-cause-of-alzheimers-might-be-coming-from-within-your-mouth?utm_source=flipboard&utm_medium=activitypub

This https://arxiv.org/abs/2503.00555 has been replaced.
initial toot: https://mastoxiv.page/@arXiv_csCR_…

Safety Tax: Safety Alignment Makes Your Large Reasoning Models Less Reasonable
Safety alignment is an important procedure before the official deployment of a Large Language Model (LLM). While safety alignment has been extensively studied for LLM, there is still a large research gap for Large Reasoning Models (LRMs) that equip with improved reasoning capability. We in this paper systematically examine a simplified pipeline for producing safety aligned LRMs. With our evaluation of various LRMs, we deliver two main findings: i) Safety alignment can be done upon the LRM to re…

Another of my forays into AI ethics is just out! This time the focus is on the ethics (or lack thereof) of Reinforcement Learning Feedback (RLF) techniques aimed at increasing the 'alignment' of LLMs.
The paper is fruit of the joint work of a great team of collaborators, among whom @… and @…

Helpful, harmless, honest? Sociotechnical limits of AI alignment and safety through Reinforcement Learning from Human Feedback - Ethics and Information Technology
This paper critically evaluates the attempts to align Artificial Intelligence (AI) systems, especially Large Language Models (LLMs), with human values and intentions through Reinforcement Learning from Feedback methods, involving either human feedback (RLHF) or AI feedback (RLAIF). Specifically, we show the shortcomings of the broadly pursued alignment goals of honesty, harmlessness, and helpfulness. Through a multidisciplinary sociotechnical critique, we examine both the theoretical underpinni…

Um homem e uma mulher, ambos sem os dois braços, decidiram casar, e
algum tempo depois tiveram filho. Qual e o nome do filme ?
Ninguém segura este bebe

A good counter-explanation from @…, although it’s not entirely consistent with the data: if this were merely aging vehicles, we’d expect to see a slow, steady decline. But we don’t; we see a •very• sharp turnaround that coincides precisely with Musk’s fascism entering full public view, and not with any sort of sudden aging event in the vehicles.
But yes, surely normal depreciation is a contributing factor as well.
https://mstdn.ca/@alessandro/114637425697204963

Why LLM Safety Guardrails Collapse After Fine-tuning: A Similarity Analysis Between Alignment and Fine-tuning Datasets
Lei Hsiung, Tianyu Pang, Yung-Chen Tang, Linyue Song, Tsung-Yi Ho, Pin-Yu Chen, Yaoqing Yang
https://arxiv.org/abs/2506.05346

Why LLM Safety Guardrails Collapse After Fine-tuning: A Similarity Analysis Between Alignment and Fine-tuning Datasets
Recent advancements in large language models (LLMs) have underscored their vulnerability to safety alignment jailbreaks, particularly when subjected to downstream fine-tuning. However, existing mitigation strategies primarily focus on reactively addressing jailbreak incidents after safety guardrails have been compromised, removing harmful gradients during fine-tuning, or continuously reinforcing safety alignment throughout fine-tuning. As such, they tend to overlook a critical upstream factor: …