Chi‐Man Vong

A Judge Should Know What Changed:Construct Validity for LLM-as-a-Judge Evaluation

Jianlin Chen, Wenhui Chen, Ziyao Lin, Chi‐Man Vong

LLM-as-a-judge evaluation is usually assessed by agreement and robustness to surface perturbations, but reliability does not establish construct validity. We formalize construct validity for an evaluator as a two-dimensional profile: invariance S, the probability that a verdict is unchanged under construct-preserving edits, …

The Free-Recipe Limit: research artefact (pre-registrations, analysis code, per-run results, and derived verdicts)

Wenhui Chen, Jianlin Chen, Ziyao Lin, Chi‐Man Vong

Frozen artefact for the paper The Free-Recipe Limit: every recipe effect measures which premise of an idealised learner broke. It contains the pre-registrations (each frozen in the job script or markdown note before the run it governs), the full analysis tree, every …

mo, cn (code pays fourni par la source)

The Free-Recipe Limit: research artefact (pre-registrations, analysis code, per-run results, and derived verdicts)

Wenhui Chen, Jianlin Chen, Ziyao Lin, Chi‐Man Vong

Frozen artefact for the paper The Free-Recipe Limit: every recipe effect measures which premise of an idealised learner broke. It contains the pre-registrations (each frozen in the job script or markdown note before the run it governs), the full analysis tree, every …

mo, cn (code pays fourni par la source)

Pixel Ignores, Superpixel Sees: Adverse Weather Image Restoration via Semantic-Center SSM

Dayu Li, Shihao Zhou, Leizhi Shu, Jin Wu et autres

Adverse weather image restoration aims to recover clear visibility from degraded images in complex weather conditions. Existing works attempt to address this problem by modeling relationships between pixels, however, this paradigm defies the spatially non-uniformity fact of degradations and learns non-discriminative features …

cn, mo (code pays fourni par la source)

Judging Is Not Enumerating: Silent Omissions in LLM-Authored Acceptable Sets

Wenhui Chen, Jianlin Chen, Ziyao Lin, Peiji Long et autres

Language models are increasingly promoted from examinees to examiners: they write the test suites, answer keys, rubrics, and reward functions that define correctness for other systems. We measure the capability that role assumes and find it lacking under the protocol the role …

Judging Is Not Enumerating: Silent Omissions in LLM-Authored Acceptable Sets

Wenhui Chen, Jianlin Chen, Ziyao Lin, Peiji Long et autres

Language models are increasingly promoted from examinees to examiners: they write the test suites, answer keys, rubrics, and reward functions that define correctness for other systems. We measure the capability that role assumes and find it lacking under the protocol the role …

mo, cn (code pays fourni par la source)

The Capability Convergence Hypothesis: Capability from Access Structure, Not Scale

Wenhui Chen, Jianlin Chen, Ziyao Lin, Chi‐Man Vong

The Platonic Representation Hypothesis (PRH) holds that as models scale, representations of heterogeneous networks converge toward a shared model of reality. We propose its sequel and boundary, the Capability Convergence Hypothesis (CCH): under a fixed per-token inference budget, representational convergence does not …

The Capability Convergence Hypothesis: Capability from Access Structure, Not Scale

Wenhui Chen, Jianlin Chen, Ziyao Lin, Chi‐Man Vong

The Platonic Representation Hypothesis (PRH) holds that as models scale, representations of heterogeneous networks converge toward a shared model of reality. We propose its sequel and boundary, the Capability Convergence Hypothesis (CCH): under a fixed per-token inference budget, representational convergence does not …

mo, cn (code pays fourni par la source)

A novel interpretable fault diagnosis approach for chillers based on self-adaptive oversampling technique and tree-based multi-meta stacking ensemble method

Zhen Jia, Guoyu Yao, Zhengdong Wang, Peng Xu et autres

Deployable real-time spinal endoscopic instance segmentation with lightweight multi-scale attention mechanism

Qi Lai, Qiang Cai, JunYan Li, Chi‐Man Vong et autres

Real-time instance segmentation in spinal endoscopy is vital for identifying and protecting key anatomy, but is hampered by narrow views, specular highlights, smoke/bleeding artifacts, fuzzy boundaries, and large-scale variation. Deployment further demands accuracy, speed, and stability under small-batch (often batch-size-one) settings. We …

cn, mo (code pays fourni par la source)

AIoU: An Adaptive Bounding Box Loss with an Edge Alignment Penalty

Wenhui Chen, Ziyao Lin, Xinyu Jiang, Chi‐Man Vong

Intersection-over-Union (IoU)-based losses, such as IoU and CIoU, have become the de facto standard for bounding box regression in modern object detectors. However, these losses mainly emphasize center distance and aspect ratio consistency, while providing limited supervision on explicit boundary alignment, which …

mo (code pays fourni par la source)

Guiding multimodal LLMs for efficient visual place recognition

Zhijian He, W. Li, Jintao Cheng, Yipu Zhang et autres

cn, hk, mo (code pays fourni par la source)