고정된 트윗
New preprint alert! We often prompt ICL tasks using either demonstrations or instructions. How much does the form of the prompt matter to the task representation formed by a language model? Stick around to find out 1/N
English
Guy Davidson
1K posts
Guy Davidson
@guyd33
Machine learning researcher @JaneStreetGroup. PhD @NYUDataScience in AI & CogSci, specifically in goals and their representations in minds & machines (he/him).
New York, USA 가입일 Nisan 2019
1.9K 팔로잉1.3K 팔로워
@eyalFeder כל הסיפור מהמם, אבל הטענה שאין שווארמה מעל בינונית בניו יורק קצת מפוקפקת… היית ב-OMG על השביעית ורחוב עשר או שבזי באמסטרדם ו-93?
(אני מניח שזה למטרות הסיפור, אבל אני גם תמיד בעד להרים לשווארמה מקומית)
עברית
האישה הנחמדה יצאה מהדוכן ואישית הובילה אותי דרך שורה של דלתות ומסדרונות שבסופן הופעתי מחדש בדיוטי פרי.
הודתי למות הדרך שלי ופניתי לאכול שווארמה בינונית (משום מה אין שווארמה בניו יורק) ולחכות לטיסה.
וכך הסתיימה החוויה החד פעמית הזאת. >>
עברית
השבוע עשיתי משהו שמעט מאוד אנשים זוכים לחוות בימי חייהם.
עשיתי קונקשן בנתב"ג.
<<
עברית
Guy Davidson 리트윗함
If “getting started with agents” feels like setup hell — same.
So we made a starter tutorial:
First agent running in <14 minutes, no Docker/AWS.
Laptop + API key only.
👇 youtube.com/watch?v=gzNW_L…
YouTube
English
@sarahcat21 I almost brought an aeropress and coffee from home before I decided that’s a bit extra. I slightly regret the decision.
English
Last time I attended NeurIPS in SoCal, I hopped a fence to get into Uber’s after party so I could check out the ice luge. Those days are gone; now I travel with a senior citizens survival kit.
English
@adinamwilliams @LakeBrenden @todd_gureckis @jcyhc_ai will present SAGE-Eval, our (w/ @LakeBrenden) systematic generalization safety benchmark at poster #1104 on Friday AM (11-2).
John does fantastic work and he's open to RE/RS roles or PhD positions in AI Safety. If you're hiring, talk to him!
x.com/jcyhc_ai/statu…
John (Yueh-Han) Chen@jcyhc_ai
Do LLMs show systematic generalization of safety facts to novel scenarios? Introducing our work SAGE-Eval, a benchmark consisting of 100+ safety facts and 10k+ scenarios to test this! - Claude-3.7-Sonnet passes only 57% of facts evaluated - o1 and o3-mini passed <45%! 🧵
English
We're also presenting some work! Our (@adinamwilliams @LakeBrenden @todd_gureckis ) interpretability work on task representations from different prompting forms will be poster #1016 on Friday's afternoon session (4:30-7:30, hall C/D/E)
x.com/guyd33/status/…
Guy Davidson@guyd33
New preprint alert! We often prompt ICL tasks using either demonstrations or instructions. How much does the form of the prompt matter to the task representation formed by a language model? Stick around to find out 1/N
English
Like ~everyone, I'll also be at #NeurIPS this week! Please reach out to chat about past (goal representations, cognitive science, intrep) or current interests (LLM mental state inference, social environments for RL). Also if you have leads on great coffee, craft beer, or tacos.
English
@redtachyon Eh my read is more of a tongue in cheek “if you know you know” than trying and failing
English
Guy Davidson 리트윗함
Stop by the Meta booth tomorrow, Wednesday Dec 3rd at #NeurIPS in San Diego! 🤖📱
We demo our new research environment, OpenApps, for digital agents. Generate thousands of app versions to train and evaluate multimodal agents to use apps like humans do.
Not attending? Stay tuned
English
@marikgoldstein I went to Modern Times Coffee nearby today, quite nice + solid breakfast tacos
English
so far in this thread i've seen
- bird rock coffee roasters
- jaunt coffee roasters
- rikka fika
- dark horse
- goldchild
thanks all!
English
what are the good* espresso places in san diego?
*wide definition of "good" here: doesn't have to be pristine, concrete counters, jazz/lofi, light roast, let-you smell-the-ground-beans-before-serving-you, etc (and better if it is not! but will tolerate it)
English
@dianarycai I got here through the conference website, tried to upload a form tonight and we'll see if it works tomorrow: neurips.myprintdesk.net
English
Where are folks going to get large posters printed near the San Diego convention center? #NeurIPS2025
English
@joannejang Absolutely, interesting and hard problem. Unclear what exactly to measure, how much of what good EQ looks like is user-dependent, and how aligned writing style/tone is with EQ (and/or the perception of it)
English
too many people conflating model eq & model personality
+
too few people who care enough about eq to work on such a fuzzy problem when there are problems far easier to measure & hill-climb on
English
Guy Davidson 리트윗함
Other great humans you might end up working with include @bvp22294, @AnsongNi, and @real_asli (in addition to several twitter-less folks).
Feel free to reach out with any questions! (though it may take me a bit to reply)
English
My team at FAIR at Meta is recruiting interns for next summer! If you're a PhD student interested in questions around theory of mind in language models for social, multi-agent settings, and have relevant background and/or experience: metacareers.com/jobs/182171308…
English
Belated update #2: my year at FAIR @AIatMeta through the AIM program was so nice that I’m sticking around for the long haul.
I’m excited to stay at FAIR and work with @real_asli and friends on fun LLM questions; I’ll be working from the New York office so we’re sticking around.
English
Belated update #1: I defended my PhD about a month ago! I appreciate the warm reception from everyone who made it in-person and virtually. Thanks to my committee, @LerrelPinto, @togelius, and Mark Ho for your feedback and fun questions.
English