CS @stanford. Reasoning enjoyer @stanfordnlp, (PO)MDPs @SISLaboratory, and speech technologies @CarnegieMellon. AGI and Emacs are cool.jemoka.com Stanford, CAJoined July 2024
A personal update:
- I just finished my 6-month residency at @AISecurityInst.
- I'm going back to MIT for the final year of my PhD.
- I'm on the postdoc and faculty job markets this fall!
Our paper on Long-term Active Embodied Question Answering (LA-EQA), in collaboration with Field AI, is accepted at #CoRL2025!
Robots that can remember, explore, and reason over days, weeks, and months.
📄Paper:arxiv.org/pdf/2507.12846
🌐Project:mind-palace-laeqa.github.io@fieldai_
We're excited to share our work on multi-objective, RL-based red-teaming! In this case, we optimize for likelihood and toxicity induction, but the algorithm presented can be used with any set of objectives. We'd love to discuss how it applies to your use cases!
We're excited to share our work on multi-objective, RL-based red-teaming! In this case, we optimize for likelihood and toxicity induction, but the algorithm presented can be used with any set of objectives. We'd love to discuss how it applies to your use cases!
New Paper Day! For EMNLP findings—in LM red-teaming, we show you have to optimize for **both** perplexity and toxicity for high-probability, hard to filter, and natural attacks!
The Transformer is not quite a Ship of Theseus, but, since 2017, best practices have radically changed for positional encodings, layer norms, attention, residual streams, and MLP components. Architecture work is not dead, and scaling is not all you need to respect scaling laws.
If you thought referencing past chats was cool, we built an MCP that lets Claude use *anything you see or do on your computer* as context. https://t.co/2hQAMmWrLM
164 Followers 1K FollowingBy real analysis, distance converge dens move position, see the problem, change the filter , Problem Define und solver| Exploitation. (optimization) Exploration
12 Followers 293 FollowingZeqi Technology (HeFei) Co., Ltd., its CEO, enterprise consulting services, currently serves more than 100 SMEs.100 small and medium-sized enterprises.
191 Followers 845 FollowingUnwitting Red Team Lead for the American Empire
Unofficial account for Paul E Lowndes:
https://t.co/nAWUtWCTb8
Officially:
https://t.co/gXgnkJLPWU
8K Followers 990 FollowingPartner @Neo. Previously @stripe, @twitter, and founder of @lightwellpro, @hullabalu, @weardrobe. Fan of startups, sci fi, and soft serve. 👩🏻💻🍦🖖🚀🇺🇸
5K Followers 692 FollowingResearch Scientist @allen_ai, PhD in NLP 🤖 UofA. Ex @GoogleDeepMind @MSFTResearch @MilaQuebec 🚨🚨 NEW BLOG about LLMs reasoning: https://t.co/Ox0iOaqY7e
9K Followers 10 FollowingFast, light, simple Docker and Linux on your Mac. Say goodbye to slow, clunky containers and VMs. Built by @kdrag0n 🚀 https://t.co/s9UtR9vmHI (hiring! DM us)
8K Followers 990 FollowingPartner @Neo. Previously @stripe, @twitter, and founder of @lightwellpro, @hullabalu, @weardrobe. Fan of startups, sci fi, and soft serve. 👩🏻💻🍦🖖🚀🇺🇸
8K Followers 198 FollowingAssistant Prof at Stanford CS, member of @stanfordnlp and statsml groups; Formerly at Microsoft / postdoc at Stanford CS / Stats.
5K Followers 692 FollowingResearch Scientist @allen_ai, PhD in NLP 🤖 UofA. Ex @GoogleDeepMind @MSFTResearch @MilaQuebec 🚨🚨 NEW BLOG about LLMs reasoning: https://t.co/Ox0iOaqY7e
1.9M Followers 27K FollowingYes, I can see some risk that your threat to jail Internet company executives for not censorsing aggressively enough could backfire.
488K Followers 146 FollowingNobel Laureate. Co-Founder & CEO @GoogleDeepMind - working on AGI. Solving disease @IsomorphicLabs. Trying to understand the fundamental nature of reality.
3K Followers 569 Followingengineer, musician, pilot, wannabe lawyer and accountant, form appreciator
prev @neuralink @hackclub @replit and open source https://t.co/bcJ8af2oEt