Tootfinder

Chain-of-Experts: Unlocking the Communication Power of Mixture-of-Experts Models
Zihan Wang, Rui Pan, Jiarui Yao, Robert Csordas, Linjie Li, Lu Yin, Jiajun Wu, Tong Zhang, Manling Li, Shiwei Liu
https://arxiv.org/abs/2506.18945

Chain-of-Experts: Unlocking the Communication Power of Mixture-of-Experts Models
We propose Chain-of-Experts (CoE), a new Mixture-of-Experts (MoE) architecture that introduces sequential expert communication within each layer. Unlike traditional MoE models, where experts operate independently in parallel, CoE processes tokens iteratively across a chain of experts inside a layer. To support dynamic expert selection across iterations, CoE employs a dedicated router at each iteration step within a layer. This design allows tokens to re-evaluate and select different experts dur…

Iterated jump noncomputability and compactness
Gavin Dooley
https://arxiv.org/abs/2506.20620 https://arxiv.org/pdf/2506.20620

Iterated jump noncomputability and compactness
We use reverse mathematics to analyze "iterated jump" versions of the following four principles: the atomic model theorem with subenumerable types (AST), the diagonally noncomputable principle (DNR), weak weak Kőnig's lemma (WWKL), and weak Kőnig's lemma (WKL). The logical relationships between these principles are summarized in Figure 1 and include, among other things, an infinite chain and an infinite antichain, the latter of which represents a strong form of non-linearity in terms of prova…

DiLQR: Differentiable Iterative Linear Quadratic Regulator via Implicit Differentiation
Shuyuan Wang, Philip D. Loewen, Michael Forbes, Bhushan Gopaluni, Wei Pan
https://arxiv.org/abs/2506.17473

DiLQR: Differentiable Iterative Linear Quadratic Regulator via Implicit Differentiation
While differentiable control has emerged as a powerful paradigm combining model-free flexibility with model-based efficiency, the iterative Linear Quadratic Regulator (iLQR) remains underexplored as a differentiable component. The scalability of differentiating through extended iterations and horizons poses significant challenges, hindering iLQR from being an effective differentiable controller. This paper introduces DiLQR, a framework that facilitates differentiation through iLQR, allowing it …

“WCAG 3, increment or overhaul?”
https://alastairc.uk/2025/06/increment-or-overhaul-wcag-3/
Taking off his hat as the AGWG chair to discuss iterating in WCAG2.
Related to the WCAG3 discussion (with notes from meeting yesterday):

Talangutveckling del 2 - Specialisering och multidrottande | Podcast https://tyngre.se/podcast/tyngre-traningssnack/talangutveckling-del-2-specialisering-och-multidrottande

Talangutveckling del 2 - Specialisering och multidrottande | Podcast
Debatten om hur unga idrottare bäst utvecklas har tagit fart i Sverige de senaste åren. I detta andra avsnitt av vår serie om talangutveckling dyker vi ner i fr

Reducible Iterated Graph Systems: multiscale-freeness and multifractals
Nero Ziyu Li, Frank Xin Hu, Thomas Britz
https://arxiv.org/abs/2506.18073 https://

Reducible Iterated Graph Systems: multiscale-freeness and multifractals
Iterated Graph Systems (IGS) aims to transplant ideas from fractal geometry into graph theory. Building on this framework, we extend Edge IGS from the primitive to the reducible setting. Within this broader context, we formulate rigorous definitions of multifractality and multiscale-freeness for graph fractals, and we establish conditions that are equivalent to the occurrence of these two phenomena. We further determine the corresponding fractal and degree spectra, proving that both are finite …

Fields with Lie-commuting and iterative operators
Jan Dobrowolski, Omar Leon Sanchez
https://arxiv.org/abs/2506.19489 https://arxiv.o…

Fields with Lie-commuting and iterative operators
We introduce a general framework for studying fields equipped with operators, given as co-ordinate functions of homomorphisms into a local algebra $\mathcal{D}$, satisfying various compatibility conditions that we denote by $Γ$ and call such structures $\mathcal{D}^Γ$-fields. These include Lie-commutativity of derivations and $\mathfrak g$-iterativity of (truncated) Hasse-Schmidt derivations. Our main result is about the existence of principal realisations of $\mathcal{D}^Γ$-kernels. As an a…

Monotonicity properties of hyperbolic projections in holomorphic iteration
Argyrios Christodoulou, Konstantinos Zarvalis
https://arxiv.org/abs/2506.19562 h…

Monotonicity properties of hyperbolic projections in holomorphic iteration
We consider hyperbolic projections of orbits of holomorphic self-maps of the unit disc, onto curves landing on the unit circle with a given angle. We show that under certain, necessary, assumptions, the projections exhibit monotonicity properties akin to those present in continuous dynamics. Our techniques are purely hyperbolic-geometric in nature and provide the general framework for analysing projections of arbitrary sequences onto curves.

On the Convergence Rates of Iterative Regularization Algorithms for Composite Bi-Level Optimization
Shimrit Shtern, Adeolu Taiwo
https://arxiv.org/abs/2506.16382

On the Convergence Rates of Iterative Regularization Algorithms for Composite Bi-Level Optimization
This paper investigates iterative methods for solving bi-level optimization problems where both inner and outer functions have a composite structure. We provide novel theoretical results, including the first convergence rate analysis for the Iteratively REgularized Proximal Gradient (IRE-PG) method, a variant of Solodov's algorithm. Our results establish simultaneous convergence rates for the inner and outer functions, highlighting the inherent trade-offs between their respective convergence ra…

Replaced article(s) found for cs.LG. https://arxiv.org/list/cs.LG/new
[3/4]:
- Iterative Minimax Games with Coupled Linear Constraints
Huiling Zhang, Zi Xu, Yu-Hong Dai