Jacob Andreas
@jacobandreas
Teaching computers to read. Assoc. prof @MITEECS / @MIT_CSAIL / @NLP_MIT (he/him). http://lingo.csail.mit.edu http://web.mit.edu/jda/www
👉 New preprint! Today, many the biggest challenges in LM post-training aren't just about correctness, but rather consistency & coherence across interactions. This paper tackles some of these issues by optimizing reasoning LMs for calibration rather than accuracy...
🚨New Paper!🚨 We trained reasoning LLMs to reason about what they don't know. o1-style reasoning training improves accuracy but produces overconfident models that hallucinate more. Meet RLCR: a simple RL method that trains LLMs to reason and reflect on their uncertainty --…
fun new paper training LLMs to analyze their own uncertainty and be more calibrated in their confidence! arxiv.org/abs/2507.16806
🚨New Paper!🚨 We trained reasoning LLMs to reason about what they don't know. o1-style reasoning training improves accuracy but produces overconfident models that hallucinate more. Meet RLCR: a simple RL method that trains LLMs to reason and reflect on their uncertainty --…
New paper on emergent reasoning about uncertainty in RL! It was great to move the needle a bit on an important problem - and very excited for future work in the space. It was an absolute pleasure working with @MehulDamani2 @ishapuri101 @IdanShenfeld @jacobandreas
🚨New Paper!🚨 We trained reasoning LLMs to reason about what they don't know. o1-style reasoning training improves accuracy but produces overconfident models that hallucinate more. Meet RLCR: a simple RL method that trains LLMs to reason and reflect on their uncertainty --…
🚨New Paper!🚨 We trained reasoning LLMs to reason about what they don't know. o1-style reasoning training improves accuracy but produces overconfident models that hallucinate more. Meet RLCR: a simple RL method that trains LLMs to reason and reflect on their uncertainty --…
RLCR produces reasoning LLMs that not only solve problems - but also reason about what they don't know. ✨🧠 📄Paper: arxiv.org/abs/2507.16806 w/ @ishapuri101, @StewartSlocum1, @IdanShenfeld, @LChoshen, @yoonrkim, @jacobandreas [9/N]
How do people reason while still staying coherent – as if they have an internal ‘world model’ for situations they’ve never encountered? A new paper on open-world cognition (preview at the world models workshop at #ICML2025!)
🚨 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:…
👉 New preprint on a new family of Transformer-type models whose depth scales logarithmically with sequence length. Enables: - fast training - fast decoding - large memory capacity in associative recall - strong length generalization on state tracking
Transformers: ⚡️fast to train (compute-bound), 🐌slow to decode (memory-bound). Can Transformers be optimal in both? Yes! By exploiting sequential-parallel duality. We introduce Transformer-PSM with constant time per token decode. 🧐 arxiv.org/pdf/2506.10918
Transformers: ⚡️fast to train (compute-bound), 🐌slow to decode (memory-bound). Can Transformers be optimal in both? Yes! By exploiting sequential-parallel duality. We introduce Transformer-PSM with constant time per token decode. 🧐 arxiv.org/pdf/2506.10918
@kaivu, @atticuswzf , and I were researching long horizon reasoning (with @jacobandreas). We found existing benchmarks’ hard problems often featured tricky puzzles, not tests of system understanding. So we made Breakpoint: a SWE benchmark designed to disambiguate this capability.
🚨🚨 Studying the INTERPLAY of LMs' internals and behavior? Join our @colmweb.org workshop on comprehensivly evaluating LMs. Deadline: June 23rd CfP: shorturl.at/sBomu We're excited to see your work!! See you in Montréal 🇨🇦 #nlproc #interpretability
For this week’s NLP Seminar, we are thrilled to host @jacobandreas to talk about “Just Asking Questions” When: 5/15 Thurs 11am PT Non-Stanford affiliates registration form: forms.gle/svy5q5uu7anHw7…
Excited to announce that this fall I'll be joining @jacobandreas's amazing lab at MIT for a postdoc to work on interp. for reasoning (with @ev_fedorenko 🤯 among others). Cannot wait to think more about this direction in such a dream academic context!
i just got an art grant from the council for the arts at MIT! *Tangible Dreams* will let visitors experiment and play with a physical neural network generating images real-time—by twisting knobs and switches, by reconnecting nodes together @ArtsatMIT
MIT NLP @ ICLR 2025 - catch @MehulDamani2 at poster 219, Thursday 3PM to chat about "Learning How Hard to Think: Input Adaptive Allocation of LM Computation"!
I am super excited to be presenting our work on adaptive inference -time compute at ICLR! Come chat with me on Thursday 4/24 at 3PM (Poster #219). I am also happy to chat about RL/reasoning/ RLHF/ inference scaling (DMs are open)!
New #NAACL2025 paper! 🚨 Transformer LMs are data hungry, we propose a new auxiliary loss function (TreeReg) to fix that. TreeReg takes bracketing decisions from syntax trees and turns them into orthogonality constraints on span representations. ✅ Boosts pre-training data…
✨ Big life updates ✨ - @afeyzaakyurek and I welcomed our baby! - Successfully defended my PhD and graduated from MIT 🎓 - Joined @OpenAI 🍓 Excited for what's next!
Tackling complex problems with LMs requires search/planning, but how should test-time compute be structured? Introducing Self-Steering, a new meta-reasoning framework where LMs coordinate their own inference procedures by writing code!