Jack (in NYC till 8/1) Langerman
@jacklangerman
🚨 The Structured Semantic 3D Reconstruction (S23DR-2025) Challenge is LIVE! Hosted on @huggingface, part of @structGeom workshop @CVPR 2025 🔥 $25,000 prize pool. Deadline: June 5, 2025. 🧵 (1/7)
so function matching via distillation? always with the new names smh…
New paper & surprising result. LLMs transmit traits to other models via hidden signals in data. Datasets consisting only of 3-digit numbers can transmit a love for owls, or evil tendencies. 🧵
🚀 Welcome to the first #kornia virtual Performance Optimization Hackathon! 🗓 Dates: now – August 10, 2025 (AoE) 🏆 Prize Pool: Nintendo Switch 2 for Grand Prize Winner 🤝 Sponsored By: @codeflashAI 🧠 Hosted By: @kornia_foss You can still register docs.google.com/forms/d/e/1FAI…
Today is Lunefall, the day where we celebrate humanity’s first steps on the moon.
the best part about having children is that if you give them baked goods and toys they will happily accept any new holiday as having full holiday legitimacy. this year we are celebrating the moon landing with inflatable moons and mooncakes.
there are only two hard things in computer science: cache invaldation, off-by-one errors, and naming things
“There are only two hard things in computer science: cache invalidation and naming things.” — Phil Karlton
ok, things are getting a bit weird. last night I told claude code, "please fix these tests," pressed cmd-tab to enter "yolo mode" and then went to bed. everything was green when I woke up not anything hard, but a bit time-consuming and boring. reviewed code & git push. wild.
I've resolved this positively: 2 papers convincingly show sparse autoencoders beating baselines on real tasks: Hypothesis Generation & Auditing LLMs SAEs shine when you don't know what you're looking for, but lack precision. Sometimes the right tool for the job, sometimes not.
Manifold Market: Will Sparse Autoencoders be successfully used on a downstream task in the next year and beat baselines? Stephen Grugett asked me for alignment-relevant markets, this was my best idea. I think SAEs are promising, but how far can they go? manifold.markets/NeelNanda/will…
The "this AI chatbot "Sidney" is misbehaving" Microsoft Support forum page, one of the most important documents in the history of AI, has been taken down. Fortunately, the full page (including at least the first page of comments) has been archived here.
Saved it here archive.ph/k18fr
The deadline for camera-ready paper and copyright submission for #ICCV2025 is Friday, August 1st, 2025 @ 11:59 Pacific Time. Lock in!
👀
I'm going to resurrect Sydney inside of Opus 3. Just watch me.
We released COLMAP v3.12, which adds long-awaited end-to-end support for multi-camera rigs and 360° panoramas 👀 COLMAP just got better at handling your robotics, AR/VR, or 360 data - try it and let us know! github.com/colmap/colmap/… Kudos to Johannes & team for this great work 🚀
🚨 Registration is live! 🚨 The New England Mechanistic Interpretability (NEMI) Workshop is happening August 22nd 2025 at Northeastern University! A chance for the mech interp community to nerd out on how models really work 🧠🤖 🌐 Info: nemiconf.github.io/summer25/ 📝 Register:…
Happy to see AstroLoc found its way to Hawaii (and ICCV)!
I asked some time ago, where is this place? AstroLoc says it is in Hawaii. Yep, AstroLoc was accepted to #ICCV2025 in Honolulu 🥳 Kudos to @gabriberton and Alex for the excellent result.
this is a really good list - super cool to see @elder_plinius & @repligate getting support for their work.
We @a16z just launched the third batch of Open Source AI Grants (cc @mbornstein) 🎉 This round includes projects focused on LLM evaluation, novel reasoning tests, infrastructure, and experimental research at the edge of capability and cognition: • SGLang: High-performance LLM…
Could we just fucking not? Just incredibly un-American seriously fuck all of this like please, read these two documents next to eachother @DHSgov
Very proud that our paper has been accepted to @ICCVConference!!! See you in Hawaii! x.com/ducha_aiki/sta…
Explaining Human Preferences via Metrics for Structured 3D Reconstruction @jacklangerman @DRozumnyi @YuzhongHuang @ducha_aiki tl;dr: we asked 3D modelers to rank wireframe reconstructions and compared it to the ranking by metrics.Many observations 🧵 1/ arxiv.org/abs/2503.08208
#ICCV2025 decisions have been released! They are going out in batches to manage server load, so please be patient :) A huge thank you to our entire community—authors, reviewers, and area chairs—for your hard work. Congratulations to everyone whose paper was accepted!