Log inSign up
Yonatan Belinkov
2,764 posts
user avatar
Yonatan Belinkov
@boknilev
Associate professor of computer science @TechnionLive; visiting scholar @KempnerInst 2025-2026.
belinkov.com
Joined April 2012
1,120
Following
5,725
Followers

New to X?

Sign up now to get your own personalized timeline!

Create account

By signing up, you agree to the Terms of Service and Privacy Policy, including Cookie Use.

Terms·Privacy·Cookies·Accessibility·Ads Info·© 2026 X Corp.
Don't miss what's happening
People on X are the first to know.
Log inSign up
  • Pinned
    user avatar
    Yonatan Belinkov
    @boknilev
    Jun 17
    Are you wondering if LLM interpretability results generalize, reproduce, etc.? Check out the reproducibility challenge and submit your work reproducing papers in this area:
    user avatar
    BlackboxNLP
    @BlackboxNLP
    May 28
    📣 Announcing the BlackboxNLP 2026 Reproducibility Challenge! A new track dedicated to rigorous robustness checks of NLP interpretability work - stress-testing baselines, ablations, generalizability, and evaluation.
    4.5K
  • user avatar
    Yonatan Belinkov
    @boknilev
    Oct 1, 2025
    Opportunities to join my group in fall 2026: * PhD applications direct or via @ELLISforEurope (ellis.eu/news/ellis-phd…) * Post-doc applications direct or via Azrieli @azrielifdn (azrielifoundation.org/fellows/intern…) or Zuckerman @stem_program (zuckermanstem.org/ourprograms/po…)
    42K
  • user avatar
    Yonatan Belinkov
    @boknilev
    Dec 24, 2018
    Interested in understanding neural networks for #NLProc ? Looking for reading material for your winter break? Check out our new paper, "Analysis Methods in Neural Language Processing: A Survey", to appear in TACL. Preprint: arxiv.org/abs/1812.08951 Website: boknilev.github.io/nlp-analysis-m…
  • user avatar
    Yonatan Belinkov
    @boknilev
    Jun 11, 2025
    After discussing the Llada paper today, we were wondering: is this just masked language modeling Bert style? Main differences seem to be: (a) training with varying masking budgets; (b) inference with gradual unmasking determined by confidence.
    arXiv logo
    arxiv.org
    Large Language Diffusion Models
    The capabilities of large language models (LLMs) are widely regarded as relying on autoregressive models (ARMs). We challenge this notion by introducing LLaDA, a diffusion model trained from...
    25K
  • user avatar
    Yonatan Belinkov
    @boknilev
    Oct 4, 2023
    Excited to see important work from @andyzou_jiaming, @hendrycks ..., on interpreting & controlling language models at representation level, to improve fairness & safety of LMs. Unfortunately it fails to engage with a large body of work on these topics from the past ~5 years.
    user avatar
    Dan Hendrycks
    @hendrycks
    Oct 3, 2023
    AI models are not just black boxes or giant inscrutable matrices. We discover they have interpretable internal representations, and we control these to influence hallucinations, bias, harmfulness, and whether a LLM lies. 🌐: ai-transparency.org 📄: arxiv.org/abs/2310.01405
    43K
  • user avatar
    Yonatan Belinkov
    @boknilev
    Jun 4, 2019
    Excited to share that I'll be joining the @TechnionLive as a Senior Lecturer (= Assistant Professor) in Computer Science in Fall 2020. Looking forward to joining new and old colleagues and working with great people on #NLProc, machine learning, and related areas.
  • user avatar
    Yonatan Belinkov
    @boknilev
    Feb 4, 2021
    #NLProc twitter: I'm teaching a seminar on interpretability and robustness in NLP next semester, and looking for key references about robustness. What are important papers about robustness that you think every #NLProc student should know? Please retweet!
    GIF
  • user avatar
    Yonatan Belinkov
    @boknilev
    Sep 22, 2025
    Excited to join @KempnerInst this year! Get in touch if you're in the Boston area and want to chat about AI interpretability, robustness, interventions, safety, multi-modality, protein/DNA LMs, new architectures, multi-agent communication, or anything else you're excited about!
    user avatar
    Kempner Institute at Harvard University
    @KempnerInst
    Sep 22, 2025
    News from the #KempnerInstitute! We’re thrilled to welcome Yonatan Belinkov (expert in #NLP) and Daphna Weinshall (expert in human & machine vision) as visiting scholars for the 2025–26 academic year. 📖 Read more: bit.ly/47QkDID #AI #MachineVision @boknilev
    15K
  • user avatar
    Yonatan Belinkov
    @boknilev
    Dec 7, 2020
    New @NeurIPSConf paper: "Investigating Gender Bias in Language Models Using Causal Mediation Analysis" Link: proceedings.neurips.cc/paper/2020/fil… With @jesse_vig @sebgehr Sharon Qian @DanielNevo @yaronsinger @pmphlt
  • user avatar
    Yonatan Belinkov
    @boknilev
    Feb 14, 2020
    Food for thought for #acl2020nlp reviewers: if the work seems "trivial", "expected", or "straightforward", this isn't necessarily a bad thing. In fact, it may mean that the authors did a good and convincing job. @pmphlt has a nice take on this: eecs.harvard.edu/~shieber/Blog/…
  • user avatar
    Yonatan Belinkov
    @boknilev
    Jun 16, 2020
    How do different design choices impact contextual word representations (CWRs)? - Building blocks: RNNs, Transformers - Objective function: autoregressive, uni/bi, masked, permutation-based - Size: width, depth
  • user avatar
    Yonatan Belinkov
    @boknilev
    Aug 16, 2023
    Am interesting workshop on language models and transformers at the @SimonsInstitute this week. All talks are on YouTube. simons.berkeley.edu/workshops/larg… I’ll talk about localization in language models tomorrow.
    8.8K
  • user avatar
    Yonatan Belinkov
    @boknilev
    Nov 28, 2020
    The unspoken law of modern NLP is that everything there is to know has already been done in Collobert and Weston 2011. I’m only half joking here.
    user avatar
    Kyunghyun Cho
    @kchonyc
    Nov 28, 2020
    Replying to @ethancaballero @Tim_Dettmers and 2 others
    i think they (and i) should! see S4.2 of jmlr.org/papers/volume1… by Collobert, @jaseweston & co. and more recently S2 of arxiv.org/abs/2004.02644 by Martins, Marinho, @andre_t_martins
  • user avatar
    Yonatan Belinkov
    @boknilev
    Feb 26, 2021
    My recent musings on (aka: a brief critical review of) probing classifiers, their promises, their shortcomings, and some alternatives: arxiv.org/abs/2102.12452 Comments welcome.
    arXiv logo
    arxiv.org
    Probing Classifiers: Promises, Shortcomings, and Advances
    Probing classifiers have emerged as one of the prominent methodologies for interpreting and analyzing deep neural network models of natural language processing. The basic idea is simple -- a...