Jacob Andreas

@jacobandreas

Teaching computers to read. Assoc. prof @MITEECS / @MIT_CSAIL / @NLP_MIT (he/him). lingo.csail.mit.edu web.mit.edu/jda/www

Cambridge, MA

Joined March 2007

Pinned
Jacob Andreas
@jacobandreas
Jul 3
👉 New preprint (we had a big backlog 😅)! Revisiting adversarial imitation learning for the era of RLVR:
Mehul Damani
@MehulDamani2
Jul 3
Higher benchmark scores do not always mean better models for users. Why? We claim that RL teaches LMs to be correct but not how to be correct: code can pass tests but be unreadable; explanations can be right but unclear. How do we train LMs to be right in the right way? (1/n)
Jacob Andreas
@jacobandreas
Jul 24
👉 New preprint! (and a direction I'm excited to do more work on soon)
Elinor
@elinorpd_
Jul 23
Pluralistic alignment is thriving as a research agenda yet failing at its goal: making the AI systems people actually use more pluralistic🌈 🚨New position paper: we argue adoption in deployed models should be the fields main goal & we provide a roadmap of how to get there 🧵1/
Jacob Andreas
@jacobandreas
Jul 7
If you're excited about Anthropic's J-space work, definitely worth checking out the original paper on Jacobian lenses by @evanqed and @arnab_api!
arxiv.org
Linearity of Relation Decoding in Transformer Language Models
Much of the knowledge encoded in transformer language models (LMs) may be expressed in terms of relations: relations between words and their synonyms, entities and their attributes, etc. We show...
Jacob Andreas
@jacobandreas
Jul 1
👉 Preprint: understanding learning dynamics & mechanisms in LMs trained to explain / predict their own behaviors!
Carl Guo
@CarlGuo866
Jul 1
New Paper 📄: LMs just want to explain themselves! When we SFT an LM on explanations of its own behaviors, do they learn to actually introspect, or do they merely imitate the original training distribution? We find evidence for the former. Despite training on a static set of
Jacob Andreas
@jacobandreas
Jun 29
👉 New preprint! Automated interpretability by approximating / replacing NN components (here attention heads) with programs.
Amiri Hayes
@amirihayes_
Jun 29
What if attention were code? We show that many attention heads in transformer LMs can be replaced by human-readable Python programs. Swap them in and the model barely notices. See our experiments here: Explaining Attention with Program Synthesis [arxiv.org/abs/2606.19317]