J Rosser
@jrosseruk
Followers
18
Following
11
Statuses
10
AI Research Lead | DPhil in Machine Learning @ Oxford
Joined April 2022
✨New Pre-Print✨ Really excited to share mine and @j_foerst new paper on LLM jailbreaking, defense and multi-agent scaffold safety with the world!
1
0
3
Some interesting results: ✅ Scaffolding appears to have less of an impact on better models ✅ It seems way easier to find unsafe scaffolds than safe ones ✅ Scaffolds reward hack, a lot More research is needed to validate these tho^
0
0
1
Had an absolute blast contributing to the very first version of Proxy! So proud of what the team have achieved, go check it out 🚀
We’re excited to announce that Proxy, Europe’s answer to @OpenAI's Operator, will be launching globally today with basic access for free and a $20p/m unlimited session plan.
1
0
3
RT @AlexDGoldie: 🇨🇦 I'll be at #NeurIPS2024 next week to present our spotlight paper on learned optimisation for RL! If you're interested i…
0
3
0
RT @AIatMeta: As we continue to explore new post-training techniques, today we're releasing Llama 3.3 — a new open source model that delive…
0
514
0
RT @yi_xin_dong: 🚀✨Introducing XGrammar: a fast, flexible, and portable engine for structured generation! 🤖Accurate JSON/grammar generatio…
0
65
0
RT @CamelAIOrg: 1st place in the 🏆 "Agents with frontier models innovation" track with @MistralAI was... Symposion: Your Research Proposal…
0
2
0