paper-with-me

홈 › Papers

Representational Depth of Evaluation Awareness Shifts With Scale in Open-Weight Language Models

2026-06-28 · Archit Manek arxiv

Do language models know when they are being tested? This question matters for AI safety: a model that recognises an evaluation context could alter its behaviour strategically, making downstream benchmarks harder to interpret. Using 11 models spanning Qwen 2.5, Gemma 2, and Llama 3.2, we find a systematic size-dependent shift in representational depth: in both Qwen 2.5 and Gemma 2, the layer at which evaluation-awareness is most linearly recoverable moves from late layers in smaller models to early layers in larger ones. This suggests that scale changes not only the strength of evaluation-awareness but also where it is most linearly recoverable in the network. This depth shift helps explain why within-family scaling trajectories are non-monotonic or inverse rather than smooth and family-general, showing that a simple universal power-law account is not supported under denser within-family sampling. Finally, white-box probe signals are consistently stronger than black-box behavioural expression, and the relationship between the two varies by family in ways not predicted by probe AUROC alone.

📄 PDF Abstract BibTeX arXiv:2606.29196

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Hierarchical Awareness Adapters with Hybrid Pyramid Feature Fusion for Dense Depth Prediction

2026-04-03 · Wuqi Su, Huilun Song, Chen Zhao, Chi Xu arxiv

Monocular depth estimation from a single RGB image remains a fundamental challenge in computer vision due to inherent scale ambiguity and the absence of explicit geometric cues. Existing approaches typically rely on incr…

Monocular Depth Estimation

Stable Diffusion Exposed: Gender Bias from Prompt to Image

2023-12-05 · Yankun Wu, Yuta Nakashima, Noa Garcia

Several studies have raised awareness about social biases in image generative models, demonstrating their predisposition towards stereotypes and imbalances. This paper contributes to this growing body of research by intr…

Image Generation

DepthFocus: Controllable Depth Estimation for See-Through Scenes

2025-11-21 · Junhong Min, Jimin Kim, Minwook Kim, Cheol-Hui Min 외 arxiv

Depth in the real world is rarely singular. Transmissive materials create layered ambiguities that confound conventional perception systems. Existing models remain passive; conventional approaches typically estimate stat…

Stereo Depth Estimation

RGB-D Salient Object Detection with Ubiquitous Target Awareness

2021-09-08 · Yifan Zhao, Jiawei Zhao, Jia Li, Xiaowu Chen

Conventional RGB-D salient object detection methods aim to leverage depth as complementary information to find the salient regions in both modalities. However, the salient object detection results heavily rely on the qua…

Objectobject-detectionObject DetectionRepresentation Learning+3

Evaluation Awareness in Language Models Has Limited Effect on Behaviour

2026-05-07 · Amelie Knecht, Lucas Florin, Thilo Hagendorff arxiv

Large reasoning models (LRMs) sometimes note in their chain of thought (CoT) that they may be under evaluation. Researchers worry that this verbalised evaluation awareness (VEA) causes models to adapt their outputs strat…