🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Super cool paper! It formalizes a lot of ideas I have been mulling over the past year, and connects so many historical ideas neatly. Definitely worth a read if you are working/interested in mechanistic interp and neural representations.

𝐇𝐨𝐰 𝐜𝐚𝐧 𝐰𝐞 𝐩𝐞𝐫𝐟𝐞𝐜𝐭𝐥𝐲 𝐞𝐫𝐚𝐬𝐞 𝐜𝐨𝐧𝐜𝐞𝐩𝐭𝐬 𝐟𝐫𝐨𝐦 𝐋𝐋𝐌𝐬? Our method, Perfect Erasure Functions (PEF), erases concepts from LLM representations w/o parameter estimation, achieving pareto optimal erasure-utility tradeoff w/ guarantees. #AISTATS2025 🧵

Training VLMs using SSL encoders? With scale you can get better performance than CLIP! Checkout our latest work led by our amazing David Fan & Peter Tong !

𝐇𝐨𝐰 𝐜𝐚𝐧 𝐰𝐞 𝐩𝐞𝐫𝐟𝐞𝐜𝐭𝐥𝐲 𝐮𝐧𝐥𝐞𝐚𝐫𝐧 𝐝𝐚𝐭𝐚 𝐟𝐫𝐨𝐦 𝐋𝐋𝐌𝐬 𝐰𝐡𝐢𝐥𝐞 𝐩𝐫𝐨𝐯𝐢𝐝𝐢𝐧𝐠 𝐠𝐮𝐚𝐫𝐚𝐧𝐭𝐞𝐞𝐬? We present S³T, a scalable unlearning framework that guarantees data deletion from LLMs by leveraging parameter-efficient fine-tuning. #ICLR2025 🧵

I’ve done way better jobs as an AC than what we got for ICML… esp. in borderline cases, I personally spend hours reading the papers and reviews in detail, overriding mediocre or wrong reviews if necessary. Papers are the lifeline of early-stage scientists, and I wouldn’t dare

📢 [Job Alert] We're hiring an MRI Analyst at the BBRC Neuroimaging Facility team in Barcelona! 🧠 Interested in neuroimaging and brain research? Apply now 👉 lnkd.in/dXrCxqpP Questions? Feel free to DM me! 💼 Fundación Pasqual Maragall

Introducing soarXiv ✈️, the most beautiful way to explore human knowledge Take any paper's URL and replace arxiv with soarxiv (show in video) to teleport to its place in the universe I've embedded all 2.8M papers up until April 2025 Try it at: soarxiv dot org

I am excited to share our #CVPR2025 paper on combining two topics close to my heart: equivariance with canonicalization and state space models. In this work, we provide a recipe to make VisionMamba equivariant/invariant by canonicalizing the patch token representations before

Fantastic work on Multi-agent RL from Dane Malenfant and Blake Richards! 🤩

New preprint! 🧠🤖 How do we build neural decoders that are: ⚡️ fast enough for real-time use 🎯 accurate across diverse tasks 🌍 generalizable to new sessions, subjects, and species? We present POSSM, a hybrid SSM architecture that optimizes for all three of these axes! 🧵1/7

💯💯 Couldn't agree more with Randall's take!

Super happy and proud to share our novel scalable RNN model - the MesaNet! This work builds upon beautiful ideas of 𝗹𝗼𝗰𝗮𝗹𝗹𝘆 𝗼𝗽𝘁𝗶𝗺𝗮𝗹 𝘁𝗲𝘀𝘁-𝘁𝗶𝗺𝗲 𝘁𝗿𝗮𝗶𝗻𝗶𝗻𝗴 (TTT), and combines ideas of in-context learning, test-time training and mesa-optimization.