Sign in

Zachary Lipton

@zacharylipton.bsky.social
9.8K followers 199 following 79 posts

Cofounder & CTO @ Abridge, Raj Reddy Associate Prof of ML @ CMU, occasional writer, relapsing 🎷, creator of d2l.ai & approximatelycorrect.com

PostsRepliesMedia
Zachary Lipton @zacharylipton.bsky.social · 20/06/2026
After countless yrs at Google DeepMind, I've decided to join [company]. Impossible decision. Grateful to @jeffdean.bsky.social & @demishassabis.bsky.social for the mentorship & late-night scientific ruminations. GDM taught me how to eat, breathe & think. Special place. Excited for their next moves 🚀
180
Reposted by Zachary Lipton
Emily Byun @yewonbyun.bsky.social · 10/10/2025
💡Can we trust synthetic data for statistical inference? We show that synthetic data (e.g., LLM simulations) can significantly improve the performance of inference tasks. The key intuition lies in the interactions between the moment residuals of synthetic data and those of real data
2389
Zachary Lipton @zacharylipton.bsky.social · 15/09/2025
We’ll know he’s really made it when Alcatraz gets autocorrected to Alcaraz.
080
Zachary Lipton @zacharylipton.bsky.social · 06/07/2025
Small idiosyncratic language errors in otherwise clear technical papers are now a plus; assurance that I'm not wasting my time trying to read GPT spam. #neurips2025
2626
Zachary Lipton @zacharylipton.bsky.social · 30/06/2025
Spent the whole day reading unlearning papers. If any of the methods worked, I might hope to recover from the experience.
090
Zachary Lipton @zacharylipton.bsky.social · 30/05/2025
Final 5m of NBA games solved. If you foul: • opposing team gets 1 FT • & keeps possession w fresh 24 Next up, Israel/Palestine.
030
Zachary Lipton @zacharylipton.bsky.social · 10/05/2025
Google miles behind Open/Anth til overnight boom 2.5. Mosaic models underwrite $1B until boom mothballed. Deepseek turns world upside down, but only for 5min. The speed at which the winds turn here is truly wild.
2160
Zachary Lipton @zacharylipton.bsky.social · 06/05/2025
What problem were airplane designers hoping to solve when they proposed this monstrosity?
270
Zachary Lipton @zacharylipton.bsky.social · 03/05/2025
50M human hours have been spent adjusting airplane seats into upright position before takeoff, saving a grand total of 0 human lives
1160
Zachary Lipton @zacharylipton.bsky.social · 30/04/2025
Unlike OpenAI, ostensibly controlled by a powerful holding entity, here have effectively 0 board power. This saga's shaping up like the Asimov's Spacer novels, but instead of each book exploring a different ratio of humans to robotic servants, it's a different level of corporate oversight.
070
Zachary Lipton @zacharylipton.bsky.social · 30/04/2025
OpenAI folks getting creative about corporate governance. I've seen this movie before... www.theinformation.com/articles/thi...
theinformation.com
Thinking Machines Lab CEO Has Unusual Control in Andreessen-Led Deal
It’s common nowadays for founders of hot startups to hold special shares that give them more voting power than most investors. But Mira Murati, a former OpenAI chief technology officer who is now a fo...
060
Zachary Lipton @zacharylipton.bsky.social · 23/04/2025
On my way to Singapore for #iclr2025. Excited to pop my head out of AI product development to get inspired, connect w the next generation of researchers, and also for food. hmu
1100
Zachary Lipton @zacharylipton.bsky.social · 07/03/2025
Controversial opinion. It’s impossible to lease an EV in the US that’s even half as good a deal as @Tesla model 3/Y.
200
Zachary Lipton @zacharylipton.bsky.social · 28/02/2025
Plugin that automatically clicks thru "reject" to select "only strictly necessary" on the cookie pop-ups. 💵
3300
Zachary Lipton @zacharylipton.bsky.social · 25/02/2025
CTO’s the most slippery title in the world. Depending on the company, you’re either 1. A technical founder who never wanted to manage, now an IC engineer 2. A VP of Engineering 3. Leading all of EPD (+AI) 4. An influencer with no internal responsibilities
4151
Zachary Lipton @zacharylipton.bsky.social · 10/02/2025
Just as the jaws of tetragenarianism are closing in, stole this W from the jaws of impending awardless decrepitude.
190
Zachary Lipton @zacharylipton.bsky.social · 07/02/2025
Phase 1 of my research career: "ML" = rebranding of the successful subset of AI. Phase 2: "AI" = rebranding of the successful subset of ML. Phase 3: AI & ML truly starting to diverge.
3130
Zachary Lipton @zacharylipton.bsky.social · 03/02/2025
Can't find any person or AI that can explain Luka → Lakers. Only satisfying theory I can conjure: behind-the-scenes feud → "calf strain" masking a relationship strain → trade demand. But according to ChatGPT's research, Luka was blindsided by the trade.
281
Zachary Lipton @zacharylipton.bsky.social · 27/01/2025
Just another boring week in the AI capabilities landscape.
1150
Reposted by Zachary Lipton
Leqi Liu @leqiliu.bsky.social · 14/12/2024
How to **efficiently** build personalized language models **without** textual info on user preferences? Our Personalized-RLHF work: - light-weight user model - personalize all *PO alignment algorithms - strong performance on the largest personalized preference dataset arxiv.org/abs/2402.05133
1122
Reposted by Zachary Lipton
Pablo Samuel Castro @pcastr.bsky.social · 10/12/2024
First #runconference @neuripsconf.bsky.social #NeurIPS2024 was great! Will share tomorrow's deets later today, join us! @zacharylipton.bsky.social @adamjelley.bsky.social @random-steve.bsky.social
2292
Zachary Lipton @zacharylipton.bsky.social · 09/12/2024
caught red-handed
070
Zachary Lipton @zacharylipton.bsky.social · 09/12/2024
Excited to attend #NeurIPS this year. AI research has never been more impactful and yet its core identity, key initiatives, and trajectory have never been less clear. Looking forward to meeting a new generation of talent, to having wild conversations, & to reconnecting w old friends.
2670
Zachary Lipton @zacharylipton.bsky.social · 28/11/2024
Thirded
0100
Zachary Lipton @zacharylipton.bsky.social · 28/11/2024
Meta-comment. Nested quote-{dunks, replies} are a navigation nightmare. One arrives at the last reply and must read the thread per reverse chronology, clicking through each tweet in sequence to get to the origin. Would love a navigation tool that jumps to the origin, progressively zooms out.
2180
Zachary Lipton @zacharylipton.bsky.social · 28/11/2024
The parallels b/w the long COVID & chronic Lyme stories (along dimensions of symptoms, narrative, & scientific controversy (over residual inflammatory syndrome vs psychology vs persistent infection)) only deepen: www.abc.net.au/listen/progr...
abc.net.au
Why long COVID may simply be 'long infection' - ABC listen
A parliamentary inquiry found between two and twenty per cent of Australians who get the virus will develop symptoms of long COVID including brain fog and fatigue, but its exact causes remain a myster...
160
Zachary Lipton @zacharylipton.bsky.social · 27/11/2024
This morning Uber Business Comfort was cheaper than UberX. Now questioning everything I thought I knew about Uber’s pricing model & constrained optimization. Is no inequality sacred? Is math real? Are corporations not only heartless but irrational? Checking into psych ward. Send thoughts & prayers.
2370
Zachary Lipton @zacharylipton.bsky.social · 27/11/2024
What BlueSky got right that Mastodon got wrong. For the 99% of users that don’t give a FF about federation, it works exactly like old Twitter, and they never need to know. Classic Apple-epitomized design principle: Advanced features should be easy to access but, when unwanted, disappear entirely.
1235620
Reposted by Zachary Lipton
Anshul Kundaje @anshulkundaje.bsky.social · 27/11/2024
What a surprise (not!). Yet again ... poor evaluations of specialized medical LLMs result in overhyped claims relative to the base LLMs. #bioMLeval
17614
Zachary Lipton @zacharylipton.bsky.social · 26/11/2024
Medically adapted foundation models (think Med-*) turn out to be more hot air than hot stuff. Correcting for fatal flaws in evaluation, the current crop are no better on balance than generic foundation models, even on the very tasks for which benefits are claimed. arxiv.org/abs/2411.04118
arxiv.org
Medical Adaptation of Large Language and Vision-Language Models: Are We Making Progress?
Several recent works seek to develop foundation models specifically for medical applications, adapting general-purpose large language models (LLMs) and vision-language models (VLMs) via continued pret...
825857
Zachary Lipton @zacharylipton.bsky.social · 25/11/2024
Puzzled by wildly weird ASR errors on iPhones (presumably a light on-device model). Today’s greatest: “dBrief”
070
Zachary Lipton @zacharylipton.bsky.social · 24/11/2024
Too soon to declare social media wars for old ML twitter are over and BS won?
22753
Zachary Lipton @zacharylipton.bsky.social · 24/11/2024
I’d love to write a blog post ruminating on all the ways the “running a startup is just like running an academic lab” trope is wildly false, and also the ways it’s kinda true…
3412
Zachary Lipton @zacharylipton.bsky.social · 23/11/2024
Moral of the story: nobody ever wanted “Twitter but X” for any X ≠ “2018”.
4190
Zachary Lipton @zacharylipton.bsky.social · 23/11/2024
Who could have suspected fraud in b-school psychology? Utterly flabbergasted. www.theatlantic.com/magazine/arc...
theatlantic.com
The Business-School Scandal That Just Keeps Getting Bigger
The rot runs deeper than almost anyone has guessed.
4340
Zachary Lipton @zacharylipton.bsky.social · 23/11/2024
Took a beat but wow.
2550
Zachary Lipton @zacharylipton.bsky.social · 19/05/2023
This feels promising.
1100