home.social

#toppaper — Public Fediverse posts

Live and recent posts from across the Fediverse tagged #toppaper, aggregated by home.social.

fetched live
  1. NEW BIML Bibliography entry

    arxiv.org/abs/2606.03811

    AI Agents Enable Adaptive Computer Worms

    Guan, Jonas, Tom Blanchard, Hanna Foerster, Hengrui Jia, Gabriel Huang, and Nicolas Papernot

    This paper is a clarion call. Time to wake up! Papernot at his best, reminding us why Machine Learning Security is crucially important. See our blog entry Echoes of the Morris Wake-up Call of 1988: berryvilleiml.com/2026/06/03/e

    #TOPPAPER #MLsec #Agents #Attacks

    berryvilleiml.com/bibliography/

  2. NEW BIML Bibliography entry

    arxiv.org/abs/2507.19457

    GEPA: Reflective Prompt Evolution Can Outperform Reinforcement Learning

    Agrawal, Lakshya A., et al

    This is an important paper moving toward the idea of agentic harnesses. There are better ways to do this, but organizing and improving harnesses is a real thing.

    #TOPPAPER #Agents #Engineering #LLM #Harness

    berryvilleiml.com/bibliography/

  3. NEW BIML Bibliography entry

    arxiv.org/pdf/2603.28052

    Meta-Harness: End-to-End Optimization of Model Harnesses

    Lee, Yoonho, Roshen Nair, Qizheng Zhang, Kangwook Lee, Omar Khattab, and Chelsea Finn

    Harnesses for Agentic AI include perception and memory devices that allow an LLM to externalize and preserve state. This work describes iterating over a set of harnesses and finding better ones. Results are impressive.

    #TOPPAPER #Agents #Engineering #LLM

    berryvilleiml.com/bibliography/

  4. NEW BIML Bibliography entry

    arxiv.org/abs/2512.24601

    Recursive Language Models

    Alex L. Zhang, Tim Kraska, Omar Khattab

    An excellent paper describing how to extend prompt context with recursion. Simple experiments. Clear explanations. This one makes you think. Are we moving towards actual Hofstaderian strange loops?

    #TOPPAPER #Engineering #LLM

    berryvilleiml.com/references/

  5. NEW BIML Bibliography entry

    arxiv.org/abs/2602.06923v1#

    From Kepler to Newton: Inductive Biases Guide Learned World Models in Transformers

    Ziming Liu, Sophia Sanborn, Surya Ganguli, Andreas Tolias

    Representation matters and is deeply constrained by tokenization. Excellent work, clearly described with real substance.

    #TOPPAPER #Representation #Tokenization

    berryvilleiml.com/references/

  6. NEW BIML Bibliography entry

    arxiv.org/abs/2503.03150

    Position: Model Collapse Does Not Mean What You Think

    Rylan Schaeffer, Joshua Kazdan, Alvan Caleb Arulandu, Sanmi Koyejo

    We think recursive pollution is a better term than model collapse. Weak terminology leads to misunderstanding of impact. See figure 4. This is a very good paper.

    #TOPPAPER #MLsec #RecursivePollution #DataPoisoning

    berryvilleiml.com/references/

  7. NEW BIML Bibliography entry

    arxiv.org/abs/2410.04840

    Strong Model Collapse

    Elvis Dohmatob, Yunzhen Feng, Arjun Subramonian, Julia Kempe
    (NYU and META)

    Recursive pollution leads to model collapse. This view of strong model collapse describes what happens in the case of recursive data poison.
    #TOPPAPER #MLsec #Data #RecursivePollution

    berryvilleiml.com/references/

  8. NEW BIML Bibliography entry

    arxiv.org/abs/2509.16499

    A Closer Look at Model Collapse: From a Generalization-to-Memorization Perspective

    Lianghe Shi, et al

    A very nice set of references to work in model collapse. Collapsed model == lookup table (that is, no generalization). Discussion of recursive pollution as causing variance shrinkage or distribution shift.

    #TOPPAPER #MLsec #Data #RecursivePollution

    berryvilleiml.com/references/

  9. NEW BIML Bibliography entry AND NEW TOP FIVE #MLsec PAPER

    READ IT

    arxiv.org/pdf/2510.07192

    Poisoning Attacks on LLMs Require a Near-constant Number of Poison Samples

    Alexandra Souly, ... Nicholas Carlini, et al

    Excellent paper, clear and well-stated (like all Carlini papers). This result shows that recursive pollution risk is even greater than we thought. Injecting backdoors is pretty easy. The examples are a bit simplistic.

    #TOPPAPER #MLsec #Attacks #DataPoisoning

    berryvilleiml.com/references/

  10. NEW BIML Bibliography entry

    arxiv.org/abs/2510.04871

    Less is More: Recursive Reasoning with Tiny Networks

    Alexia Jolicoeur-Martineau

    This is an engineering exercise akin to “set it to 57,” but it is really interesting. A set of weekend kludges that has important implications. Harold wants to pursue this line to think about how integers are represented. Won the ARC prize.

    #TOPPAPER #Engineering #MLsec

    berryvilleiml.com/references/

  11. NEW BIML Bibliography entry

    nature.com/articles/s41586-025

    Optical generative models

    Chen, Shiqi, Yuhang Li, Yuntian Wang, Hanlong Chen and Aydogan Ozcan

    Light for computation with properties of low power and superposition. Analo of quantum computing. This reminds os of Rosenblatt’s Perceptrons from the ’50s.

    #TOPPAPER #MLsec #Engineering

    berryvilleiml.com/references/

  12. NEW BIML Bibliography entry

    direct.mit.edu/books/oa-monogr

    Chapter 13, Context Changes Everything

    Alicia Juarrero

    A solid treatment of the 4Es theory (Embodied, Embedded, Extended, Enactive) properly grounded in philosophy of mind.

    #MLsec #ML #TOPPAPER

    berryvilleiml.com/references/

  13. NEW BIML Bibliography entry

    bpb-us-w2.wpmucdn.com/web.sas.

    The Evolution of Intelligence and Access to Cognitive Unconscious

    Paul Rozin

    Social learning, tool use, and psychology. An ancient and still relevant paper. Discusses evolution and genetic programming.

    #MLsec #ML #TOPPAPER

    berryvilleiml.com/references/

  14. NEW BIML Bibliography entry

    arxiv.org/abs/2506.21521v1

    Potemkin Understanding in Large Language Models

    Marina Mancoridis et al

    Consistency checks demonstrate the paucity of LLM "concepts." Error-making in systems reveals important issues. This work rediscovers something important and introduces some new nomenclature for describing it.

    #MLsec #ML #TOPPAPER

    berryvilleiml.com/references/

  15. NEW BIML Bibliography entry

    arxiv.org/abs/2505.11581

    Questioning Representational Optimism in Deep Learning: The Fractured Entangled Representation Hypothesis

    Akarsh Kumar et al

    Excellent work on representation in the face of stochastic gradient descent. This is a top paper in representation. Introduces FER. Toward creativity and intuition.

    #MLsec #ML #TOPPAPER
    berryvilleiml.com/references/