Papers where the original authors have engaged with our plain-language explanation.
For every paper on this page, at least one of the original authors has seen our plain-language explanation and engaged with it — either confirming it reads accurately or requesting corrections that we then applied. An endorsement does not mean the authors formally approve every sentence, but it does mean the explanation has passed the eyes of the people who wrote the paper.
1363 papers reviewed by authors · 1081–1090 / 1363
This paper addresses reproducibility challenges in agent research by introducing "rollout cards," a standardized publication format that preserves raw rollout records and explicitly declares reporting rules to ensure transparent and verifiable evaluation of agentic systems.
This paper formalizes the phenomenon of divergent conclusions from identical observations as a structural issue of world-model non-identifiability arising from differing inference profiles and learning biases, rather than as a defect of rationality or good faith.
This study utilizes high-resolution observations of the Cepheus A massive protostellar cluster to demonstrate that gravity drives gas inflows while magnetic fields cooperatively regulate turbulence, revealing a coherent, multiscale alignment of gravitational, magnetic, and velocity fields that challenges the conventional view of magnetic fields as purely resistive forces in star formation.
This paper proposes a modified measure formalism where dynamical, non-universal string tensions coupled to a new tension scalar field generate induced de Sitter space on a braneworld while simultaneously predicting Dark Matter as "Dark copies" of the Standard Model with distinct tensions.
This position paper argues that toxicity detection should shift from treating toxicity as an intrinsic property of isolated text to measuring it as contextual communicative harm, introducing the Contextual Stress Framework (CSF) and CSF-Eval to better capture how perceived norm violations and social context generate actual disruption.
This paper demonstrates that for small language models (2-3B parameters), a structured 4-stage harness pipeline significantly outperforms raw prompting or minimal wrappers in operational stability and task success, while also revealing that insufficient scaffolding can lead to structural format collapse in certain models.
The paper introduces ShapeCodeBench, a renewable synthetic benchmark designed to evaluate models on perception-to-program reconstruction tasks by requiring them to generate executable drawing programs from rendered images, revealing that while current multimodal models preserve foreground structure, they still struggle with exact match due to minor parameter errors.
This paper presents a neuro-symbolic framework that integrates a GLLM generator with a Separation Logic verifier to enable the self-correcting, collision-free production of G-code by translating logical proof failures into precise spatial directives for iterative refinement.
This paper demonstrates that in the semi-classical limit, highly-excited hydrogen atom energy eigenfunctions correspond not to a single classical orbit but to an equal-weight superposition of classical elliptic orbits sharing the same energy and angular momentum, as evidenced by the matching of quantum and classical probability distributions.
This paper demonstrates that incorporating a time-dependent torsion scalar into a non-interacting holographic dark energy model with the Hubble radius as an infrared cut-off successfully drives cosmic acceleration and yields an equation of state distinct from the cosmological constant, thereby resolving limitations found in earlier interacting models.