Jacob Andreas
@jacobandreas
Followers
20K
Following
1K
Media
86
Statuses
3K
Teaching computers to read. Assoc. prof @MITEECS / @MIT_CSAIL / @NLP_MIT (he/him). https://t.co/5kCnXHjtlY https://t.co/2A3qF5vdJw
Cambridge, MA
Joined March 2007
👉 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 --
2
11
100
RT @LAW2025_NeurIPS: 📢 Thrilled to announce LAW 2025 workshop, Bridging Language, Agent, and World Models, at #NeurIPS2025 this December in….
0
10
0
RT @ishapuri101: fun new paper training LLMs to analyze their own uncertainty and be more calibrated in their confidence! .
arxiv.org
When language models (LMs) are trained via reinforcement learning (RL) to generate natural language "reasoning chains", their performance improves on a variety of difficult question answering...
0
10
0
RT @StewartSlocum1: New paper on emergent reasoning about uncertainty in RL!. It was great to move the needle a bit on an important problem….
0
2
0
RLCR helps, but is far from solving the problem. I think there's a ton of room to develop improved post-training procedures to bring LMs external expressions of certainty into alignment with their internal representations of confidence, within & across samples.
2
0
4
reducing hallucination (esp. on OOD) tasks. My favorite figure is this one, which shows the sum of the probabilities a reasoning LMs assign to mutually exclusive answers across reasoning chains. On some questions, models assign full confidence to 10+ different answers! Here.
1
0
2
RT @MehulDamani2: 🚨New Paper!🚨.We trained reasoning LLMs to reason about what they don't know. o1-style reasoning training improves accura….
0
268
0
RT @MehulDamani2: RLCR produces reasoning LLMs that not only solve problems - but also reason about what they don't know. ✨🧠.📄Paper: https:….
arxiv.org
When language models (LMs) are trained via reinforcement learning (RL) to generate natural language "reasoning chains", their performance improves on a variety of difficult question answering...
0
4
0
RT @T_BrookeWilson: How do people reason while still staying coherent – as if they have an internal ‘world model’ for situations they’ve ne….
0
26
0
RT @kpal_koyena: 🚨 Registration is live! 🚨. The New England Mechanistic Interpretability (NEMI) Workshop is happening August 22nd 2025 at N….
0
28
0
👉 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. 🧐
2
9
82
RT @MorrisYau: Transformers: ⚡️fast to train (compute-bound), 🐌slow to decode (memory-bound). Can Transformers be optimal in both? Yes! By….
0
38
0
RT @uzpg_: @kaivu, @atticuswzf , and I were researching long horizon reasoning (with @jacobandreas). We found existing benchmarks’ hard pro….
0
10
0
RT @interplaywrkshp: 🚨🚨 Studying the INTERPLAY of LMs' internals and behavior? . Join our @colmweb.org workshop on comprehensivly evaluati….
0
4
0
RT @stanfordnlp: For this week’s NLP Seminar, we are thrilled to host @jacobandreas to talk about “Just Asking Questions”.When: 5/15 Thurs….
0
12
0
RT @LauraRuis: Excited to announce that this fall I'll be joining @jacobandreas's amazing lab at MIT for a postdoc to work on interp. for r….
0
11
0
RT @cedcolas: i just got an art grant from the council for the arts at MIT!. *Tangible Dreams* will let visitors experiment and play with a….
0
12
0
RT @nlp_mit: MIT NLP @ ICLR 2025 - catch.@MehulDamani2 at poster 219, Thursday 3PM to chat about "Learning How Hard to Think: Input Adaptiv….
0
1
0
RT @ShikharMurty: New #NAACL2025 paper! 🚨.Transformer LMs are data hungry, we propose a new auxiliary loss function (TreeReg) to fix that.….
0
24
0
RT @akyurekekin: ✨ Big life updates ✨. - @afeyzaakyurek and I welcomed our baby!.- Successfully defended my PhD and graduated from MIT 🎓.-….
0
12
0