Test-Time Optimization

Follow this topic Ask about this topic

Test-time optimization studies how to improve a pretrained model's answer or behavior using information available for the current input, rather than relying only on training-time changes. Approaches include sampling and verifying multiple candidates, gradient or zeroth-order refinement of latents and outputs, constraint-based optimization, and online updates to model parameters. These methods are used in reasoning and code generation as well as 3D vision, image generation, robotics, and scientific design, where feedback may come from tests, physical constraints, or the input itself. A central challenge is getting reliable gains within a limited compute budget: weak verifiers can reward the wrong result, and optimizing for a metric may not improve real-world quality.

Latest papers

38 papers3 papers in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.

PhyMix: Towards Physically Consistent Single-Image 3D Indoor Scene Generation with Implicit--Explicit Optimization
Apr 11, 2026Dongli Wu, Jingyu Hu, Ka-Hei Hui +4Image-to-3D Scene Generation 3D Scene Generation
1Lingnan University · 2The Hong Kong University of Science and Technology (Guangzhou) · 3The Chinese University of Hong Kong +2
Learning 3D Reconstruction with Priors in Test Time
Apr 4, 2026Lei Zhou, Haoyu Wu, Akshat Dave +1Multi-View 3D Reconstruction Test-Time Optimization
Stony Brook University
Diff3R: Feed-forward 3D Gaussian Splatting with Uncertainty-aware Differentiable Optimization
Apr 1, 2026Yueh-Cheng Liu, Jozef Hladký, Matthias Nießner +1Implicit Differentiation 3D Gaussian Splatting
Technical University of Munich · Computing Systems Lab, Huawei Technologies, Switzerland
Scaling Atomistic Protein Binder Design with Generative Pretraining and Test-Time Compute
Mar 30, 2026Kieran Didi, Zuobai Zhang, Guoqing Zhou +11Protein Binder Design Protein Design
NVIDIA · University of Oxford · Mila - Québec AI Institute +8
Poppy: Polarization-based Plug-and-Play Guidance for Enhancing Monocular Normal Estimation
Mar 29, 2026Irene Kim, Sai Tanmay Reddy Chakkera, Alexandros Graikos +2Surface Normal Estimation Test-Time Optimization
Stony Brook University
Meta-Learned Adaptive Optimization for Robust Human Mesh Recovery with Uncertainty-Aware Parameter Updates
Mar 27, 2026Shaurjya Mandal, Nutan Sharma, John GaleottiHuman Mesh Recovery Test-Time Optimization
$\nabla$ -Reasoner: LLM Reasoning via Test-Time Gradient Descent in Latent Space
Mar 5, 2026Peihao Wang, Ruisi Cai, Zhen Wang +4LLM Reasoning Test-Time Optimization
The University of Texas at Austin · UC San Diego · TTIC +1
TTSR: Test-Time Self-Evolving via Reflection
Feb 6, 2026Haoyang He, Zihua Rong, Yunjia Zhao +3LLM Self-Refinement Test-Time Optimization
Beijing University of Posts and Telecommunications · Southwestern University of Finance and Economics · China Unicom Online Information Technology Co., Ltd.

Apr 11, 2026J/K move · Enter open · S save

PhyMix: Towards Physically Consistent Single-Image 3D Indoor Scene Generation with Implicit--Explicit Optimization

Dongli Wu, Jingyu Hu, Ka-Hei Hui, Xiaobao Wei+3

1Lingnan University · 2The Hong Kong University of Science and Technology (Guangzhou) · 3The Chinese University of Hong Kong+2

Image-to-3D Scene Generation 3D Scene Generation Indoor Scene Generation 3D Scene Layout Generation Physics-Informed Generative Modeling Test-Time Optimization

Open paper AskPDF arXiv

Existing methods for single-image 3D indoor scene generation often produce visually plausible results that violate real-world physical constraints, limiting their reliability in robotics, embodied AI, and virtual environment design. To bridge this gap, we introduce a Physics Evaluator that assesses four aspects (geometric priors, contact, stability, and deployability) through nine sub-constraints, establishing a benchmark for evaluating physical consistency. Our analysis reveals that even state-of-the-art methods produce scenes with substantial physical inconsistencies despite their visual plausibility. To address these limitations, we propose a framework that incorporates feedback from the Physics Evaluator into both training and inference to improve the physical plausibility of generated scenes. Specifically, PhyMix comprises two components: (i) implicit alignment via Scene-GRPO, a critic-free group-relative preference optimization strategy that uses physical evaluations to guide the generator toward physically plausible layouts, and (ii) explicit refinement via verifier-guided agentic repair, which translates evaluator diagnostics into executable scene-editing skills to correct residual violations at inference time. PhyMix unifies physical evaluation, preference-based alignment, and agentic inference-time correction to improve physical plausibility while retaining the observed scene content. Experiments show improvements in reconstruction quality and physical plausibility, while qualitative examples on stylized and real-world images illustrate the applicability of the framework beyond synthetic inputs.