🔥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 w consequences for how we understand membership, unlearning, training data memorisation, ...

I've been lucky enough to see an early draft of this. It has a surprising RL angle! The RL community has long suspected that the Decision Transformer might be doing “trajectory stitching,” but I haven’t seen empirical evidence yet. Ken’s paper shows how subsequences can be

important, timely work for the imminent gen ai “legal moment”

Really really interesting work. Adds to the pile of evidence that train-test decontamination is super hard, and we are probably not doing a good job of this in general.

Want to learn the engineering details of building state-of-the-art Large Language Models (LLMs)? Not finding much info in OpenAI’s non-technical reports? Percy Liang and Tatsunori Hashimoto are here to help with CS336: Language Modeling from Scratch, now rolling out to YouTube.

Thrilled to know that our paper, `Safety Alignment Should be Made More Than Just a Few Tokens Deep`, received the ICLR 2025 Outstanding Paper Award. We sincerely thank the ICLR committee for awarding one of this year's Outstanding Paper Awards to AI Safety / Adversarial ML.

will present this work as spotlight at icml this year :)

40% with just 1 try per task: SWE-agent-LM-32B is the new #1 open source model on SWE-bench Verified. We built it by synthesizing a ton of agentic training data from 100+ Python repos. Today we’re open-sourcing the toolkit that made it happen: SWE-smith.

truly open foundation models that anyone can contribute to!

kinda incredible

Come read about all the mistakes I made along the way to beating Llama 3.1 8B on 14/19 benchmarks. We trained from scratch, made plenty of wrong turns, and learned a lot.

AI agents have the potential to significantly alter the cybersecurity landscape. To help us understand this change, we are excited to release BountyBench, the first framework to capture offensive & defensive cyber-capabilities in evolving real-world systems.

Is AI already superhuman at FrontierMath? To answer this question, we ran a competition at MIT, pitting eight teams of mathematicians against o4-mini-medium. Result: o4-mini beat all but two teams. And while AIs aren't yet clearly superhuman, they probably will be soon.

new paper! 🫡 why are state space models (SSMs) worse than Transformers at recall over their context? this is a question about the mechanisms underlying model behaviour: therefore, we propose using mechanistic evaluations to answer it!

🚀 DeepSeek-R1-0528 is here! 🔹 Improved benchmark performance 🔹 Enhanced front-end capabilities 🔹 Reduced hallucinations 🔹 Supports JSON output & function calling ✅ Try it now: chat.deepseek.com 🔌 No change to API usage — docs here: api-docs.deepseek.com/guides/reasoni… 🔗

🎀 fine-grained, interpretable representation steering for LMs! meet RePS — Reference-free Preference Steering! 1⃣ outperforms existing methods on 2B-27B LMs, nearly matching prompting 2⃣ supports both steering and suppression (beat system prompts!) 3⃣ jailbreak-proof (1/n)

we present a new representation steering training objective to rival against prompting! and you also get: - a fun trick: you can mitigate the side-effects of randomly selecting steering factors by simply training with it. - a long appendix with our core dumps on steering