Tootfinder

Feature-Guided Diffusion for Non-Differentiable Inverse Rendering
Andrei-Timotei Ardelean, Michael Fischer, Tim Weyrich, Tom\'a\v{s} Iser
https://arxiv.org/abs/2607.17411 https://arxiv.org/pdf/2607.17411 https://arxiv.org/html/2607.17411
arXiv:2607.17411v1 Announce Type: new
Abstract: Inverse rendering is traditionally solved via differentiable renderers and gradient descent, which requires substantial problem-specific engineering and is prone to getting stuck in local minima due to ambiguities. Derivative-free approaches alleviate engineering requirements, but often heavily depend on a good problem initialization. In this work, we propose Feature-Informed Diffusion Evolution (FIDE), a fully black-box framework that requires no gradients or specific initialization: the renderer is treated as an opaque function whose only requirement is to produce images. Our key insight is feature guiding: rather than reducing each candidate rendering to a scalar loss value, we use a Vision Transformer (ViT) to extract dense visual features from it. We subsequently use these features to train a diffusion-based candidate proposal model, allowing the network to use visual cues to predict parameters that would match the target image. The candidate solutions proposed by this diffusion model are then refined in a closed loop with a CMA evolution strategy, continuously narrowing the proposal region as optimization progresses. We validate across diverse inverse problems from path tracing, vector splines, Voronoi shaders, and robotics, and demonstrate that feature-guiding substantially improves convergence over scalar-loss baselines and reliably escapes local minima where gradient-based methods stall.
toXiv_bot_toot

One of the beautiful things about Sublime Merge¹ (and git/diffs) is that you can see exactly what has changed in complex expected values in tests to ensure that you’re updating the tests without overlooking regressions.
(This is from the Markdown page loader tests in Kitten², as I’m refactoring to implement the upcoming breaking change in the stateful components API³ as it affects the generated code for stateful layout components in Markdown pages.)
¹ Which I always have running,…

Appartement avec vue sur le Nil, 30 centimes d’euro par mois : en Egypte, la fin annoncée des loyers bloqués
https://www.lemonde.fr/afrique/article/2026/04/27/apparteme…

Appartement avec vue sur le Nil, 30 centimes d’euro par mois : en Egypte, la fin annoncée des loyers bloqués
Depuis des décennies, des millions d’Egyptiens payaient des loyers dérisoires grâce à une loi qui gelait les prix. Une réforme, voulue par le président Al-Sissi, va remettre en question ce pilier du contrat social.

The ‘plug-and-play’ solar revolution
It’s a $400 kit from Bright Saver,
a non-profit advocating for
“plug-and-play” solar that works
for renters and homeowners alike.
Unlike traditional rooftop solar, which requires thousands of dollars in upfront costs, specialized mounting hardware and professional electricians,
this system is designed for the everyday consumer.
Setup is simple:
you hang the panel on a balcony or prop it up in a back yard and…

Rethinking Entropy Minimization in Test-Time Adaptation for Autoregressive Models
Wei-Ping Huang, Chee-En Yu, Guan-Ting Lin, Hung-yi Lee
https://arxiv.org/abs/2605.08186 https://arxiv.org/pdf/2605.08186 https://arxiv.org/html/2605.08186
arXiv:2605.08186v1 Announce Type: new
Abstract: Test-Time Adaptation (TTA) via entropy minimization (EM) has proven effective for classification tasks, yet its application to generative autoregressive models remains theoretically fragmented. Existing approaches typically rely on distinct heuristics, such as teacher forcing with pseudo labels or policy-gradient-based reinforcement learning, without a unified mathematical foundation. In this work, we resolve this discrepancy by deriving a rigorous formulation of EM tailored to autoregressive models. We show that the exact objective naturally decomposes into a token-level policy gradient loss and a token-level entropy loss, and we reinterpret prior methods as partial realizations of this unified formulation. Using Whisper ASR as a testbed, we demonstrate that our approach consistently improves performance across more than 20 diverse domains, including acoustic noise, accents, and multilingual settings.
toXiv_bot_toot