Soft Contamination Means Benchmarks Test Shallow Generalization https://arxiv.org/abs/2602.12413

Data Repetition Beats Data Scaling in Long-CoT Supervised Fine-Tuning https://arxiv.org/abs/2602.11149

rePIRL: Learn PRM with Inverse RL for LLM Reasoning https://arxiv.org/abs/2602.07832

Tensor Decomposition for Non-Clifford Gate Minimization https://arxiv.org/abs/2602.15285

BRIDGE: Predicting Human Task Completion Time From Model Performance https://arxiv.org/abs/2602.07267

CCBench https://ccbench.org/

Benchmarks Saturate When The Model Gets Smarter Than The Judge https://arxiv.org/abs/2601.19532

SETI@home: Data Acquisition and Front-end Processing
https://iopscience.iop.org/article/10.3847/1538-3881/ade5a7
SETI@home: Data Analysis and Findings https://iopscience.iop.org/article/10.3847/1538-3881/ade5ab

Узнал из https://t.iss.one/tropicalgeometry/1095 что оптимальность решётки E8 тоже формализовали!* Интересно почитать блюпринт (и вообще кажется, что это приятный жанр): файл содержит схему доказательства, причем доступную человеку со стороны. Ну и забавен местами (см. картинку)

https://thefundamentaltheor3m.github.io/Sphere-Packing-Lean/blueprint.pdf

You Don't Need to Run Every Eval
https://fixupx.com/DimitrisPapail/status/2026531440414925307

Charge-4e superconductor with parafermionic vortices: A path to universal topological quantum computation https://arxiv.org/abs/2602.06963

Gradient Regularization Prevents Reward Hacking in Reinforcement Learning from Human Feedback and Verifiable Rewards https://arxiv.org/abs/2602.18037

Topological superconductivity with emergent vortex lattice in twisted semiconductors https://arxiv.org/abs/2602.15106

H-Neurons: On the Existence, Impact, and Origin of Hallucination-Associated Neurons in LLMs
https://arxiv.org/abs/2512.01797
via @Futuris

Bullshit Benchmark https://github.com/petergpt/bullshit-benchmark

Scaling Diffusion Language Models via Adaptation from Autoregressive Models https://arxiv.org/abs/2410.17891

Bayesians Commit the Gambler's Fallacy

Abstract:

The gambler's fallacy is the tendency to expect random processes to switch more often than they actually do—for example, to assign a higher probability to heads after a streak of tails. It's often taken to be evidence for irrationality. It isn't. Rather, it's to be expected from a group of Bayesians who begin with causal uncertainty, and then observe unbiased data from an (in fact) statistically independent process. Although they increase their confidence that the outcomes are independent, they do so in an asymmetric way—ruling out “streaky” hypotheses more quickly than “switchy” ones. Their expectations depend on this balance of uncertainty; as a result, the majority (and the average) exhibit the gambler's fallacy, expecting a heads after a string of tails. If they have limited memory, this tendency persists even with arbitrarily-large amounts of data. In fact, such Bayesians exhibit a variety of the empirical trends found in studies of the gambler's fallacy. They expect switches after short streaks but continuations after long ones; these nonlinear expectations vary with their familiarity with the causal system; their predictions depend on the sequence they've just seen; they produce sequences that are too switchy; and they exhibit greater rates of the gambler's fallacy in binary predictions than in probability estimates. In short: what's been thought to be evidence for irrationality may instead be rational responses to limited data and memory.

https://onlinelibrary.wiley.com/doi/10.1111/cogs.70171