Had a wonderful time at #NAACL2024 this week! Thanks to everyone who came to my oral presentation on defending LLMs against backdoor attacks!

Meet Tülu 3 -- a set of state-of-the-art instruct models with fully open data, eval code, and training algorithms. We invented new methods for fine-tuning language models with RL and built upon best practices in the community to scale synthetic instruction and preference data.

Super excited to release Tülu 3, a family of fully-open state-of-the-art post-trained models, including its data, eval, code, and training recipes in a comprehensive guide for post-training techniques! allenai.org/papers/tulu-3-…

Introducing IFBench, a benchmark to measure how well AI models follow new, challenging, and diverse verifiable instructions. Top models like Gemini 2.5 Pro or Claude 4 Sonnet are only able to score up to 50%, presenting an open frontier for post-training. 🧵

This new benchmark created by Valentina Pyatkin should be the new default replacing IFEval. Some of the best frontier models get <50% and it comes with separate training prompts so people don’t effectively train on test. Wild gap from o3 to Gemini 2.5 pro of like 30 points.

Worried about overfitting to IFEval? 🤔 Use ✨IFBench✨ our new, challenging instruction-following benchmark! Loved working w/ Valentina Pyatkin! Personal highlight: our multi-turn eval setting makes it possible to isolate constraint-following from the rest of the instruction 🔍

🤔 How do we train AI models that surpass their teachers? 🚨 In #COLM2025: ✨Delta learning ✨makes LLM post-training cheap and easy – with only weak data, we beat open 8B SOTA 🤯 The secret? Learn from the *differences* in weak data pairs! 📜 arxiv.org/abs/2507.06187 🧵 below

A game-changer for post-training!