arXiv subjects

Sheng Ren

Publications and source records attributed to Sheng Ren.

6 recordsLinked to original sources

arXiv2026

Rethinking Normalization Placement for LLMs: Post-Norm under Curriculum Depth Growing

Pre-norm is the standard normalization placement in modern Transformers because it facilitates joint optimization of full-depth models. We ask whether this preference persists when depth is introduced through a curriculum. In curriculum depth growth, each appended block receives the boundary representation produced by a trained prefix, making normalization placement relevant to forward conditioning. We therefore test whether placement and training curriculum interact. In a controlled distillation study with a Qwen3-8B teacher and a nine-layer student, pre-norm and post-norm are indistinguishable under joint training, differing by $0.0004$ validation CE, while post-norm improves over pre-norm by $0.0328$ under curriculum growth, an order of magnitude larger. A post-joint control matched by student active-layer tokens remains worse than post-grow, which rules out compute as the sole explanation. The ranking crosses over during the curriculum: post-norm takes the lead once blocks are appended. Single-block and freeze controls localize the ranking change to block appending rather than shallow-block quality or retraining. Boundary diagnostics associate post-norm with stable residual scales and pre-norm with structural-token scale drift; on a fixed batch, the final pre-grow block is also nearly identity-mapped. Together with the phase-wise crossover, these observations are consistent with boundary-scale conditioning after new blocks are appended. The results motivate treating normalization placement and training curriculum as coupled design choices in this distillation setting.

cs.AI↗

arXiv2026

Compress the Easy, Explore the Hard: Difficulty-Aware Entropy Regularization for Efficient LLM Reasoning

Chain-of-Thought (CoT) has substantially empowered Large Language Models (LLMs) to tackle complex reasoning tasks, yet the verbose nature of explicit reasoning steps incurs prohibitive inference latency and computational costs, limiting real-world deployment. While existing compression methods - ranging from self-training to Reinforcement Learning (RL) with length constraints - attempt to mitigate this, they often sacrifice reasoning capability for brevity. We identify a critical failure mode in these approaches: explicitly optimizing for shorter trajectories triggers rapid entropy collapse, which prematurely shrinks the exploration space and stifles the discovery of valid reasoning paths, particularly for challenging questions requiring extensive deduction. To address this issue, we propose Compress responses for Easy questions and Explore Hard ones (CEEH), a difficulty-aware approach to RL-based efficient reasoning. CEEH dynamically assesses instance difficulty to apply selective entropy regularization: it preserves a diverse search space for currently hard questions to ensure robustness, while permitting aggressive compression on easier instances where the reasoning path is well-established. In addition, we introduce a dynamic optimal-length penalty anchored to the historically shortest correct response, which effectively counteracts entropy-induced length inflation and stabilizes the reward signal. Across six reasoning benchmarks, CEEH consistently reduces response length while maintaining accuracy comparable to the base model, and improves Pass@k relative to length-only optimization.

cs.LG↗

arXiv2025

Humanoid-inspired Causal Representation Learning for Domain Generalization

This paper proposes the Humanoid-inspired Structural Causal Model (HSCM), a novel causal framework inspired by human intelligence, designed to overcome the limitations of conventional domain generalization models. Unlike approaches that rely on statistics to capture data-label dependencies and learn distortion-invariant representations, HSCM replicates the hierarchical processing and multi-level learning of human vision systems, focusing on modeling fine-grained causal mechanisms. By disentangling and reweighting key image attributes such as color, texture, and shape, HSCM enhances generalization across diverse domains, ensuring robust performance and interpretability. Leveraging the flexibility and adaptability of human intelligence, our approach enables more effective transfer and learning in dynamic, complex environments. Through both theoretical and empirical evaluations, we demonstrate that HSCM outperforms existing domain generalization models, providing a more principled method for capturing causal relationships and improving model robustness. The code is available at https://github.com/lambett/HSCM.

cs.AI↗

arXiv2025

Clock Pulling Enables Maximum-Efficiency Wireless Power Transfer

Nonlinear parity-time (PT) symmetry in non-Hermitian wireless power transfer (WPT) systems, while attracting significant attention from both physics and engineering communities, have posed formidable theoretical and practical challenges due to their complex dynamical mechanisms. Here, we revisit multistability in nonlinear non-Hermitian systems and find that the PT-symmetry state is not always stable even in PT-symmetry phase. We report a discovery on a nonlinear clock-pulling mechanism, which can forcibly break the PT symmetry. Proper implementation of this mechanism can switch the system stability, particularly in stabilizing the conventional unstable state which has the maximum transfer efficiency for WPT. Our work offers new tools for non-Hermitian physics and is expected to drive technological progress.

physics.app-ph↗

arXiv2024

Fast Wrong-way Cycling Detection in CCTV Videos: Sparse Sampling is All You Need

Effective monitoring of unusual transportation behaviors, such as wrong-way cycling (i.e., riding a bicycle or e-bike against designated traffic flow), is crucial for optimizing law enforcement deployment and traffic planning. However, accurately recording all wrong-way cycling events is both unnecessary and infeasible in resource-constrained environments, as it requires high-resolution cameras for evidence collection and event detection. To address this challenge, we propose WWC-Predictor, a novel method for efficiently estimating the wrong-way cycling ratio, defined as the proportion of wrong-way cycling events relative to the total number of cycling movements over a given time period. The core innovation of our method lies in accurately detecting wrong-way cycling events in sparsely sampled frames using a light-weight detector, then estimating the overall ratio using an autoregressive moving average model. To evaluate the effectiveness of our method, we construct a benchmark dataset consisting of 35 minutes of video sequences with minute-level annotations.Our method achieves an average error rate of a mere 1.475\% while consuming only 19.12\% GPU time required by conventional tracking methods, validating its effectiveness in estimating the wrong-way cycling ratio. Our source code is publicly available at: https://github.com/VICA-Lab-HKUST-GZ/WWC-Predictor.

cs.CV↗

arXiv2020

Towards Class-incremental Object Detection with Nearest Mean of Exemplars

Incremental learning is a form of online learning. Incremental learning can modify the parameters and structure of the deep learning model so that the model does not forget the old knowledge while learning new knowledge. Preventing catastrophic forgetting is the most important task of incremental learning. However, the current incremental learning is often only for one type of input. For example, if the input images are of the same type, the current incremental model can learn new knowledge while not forgetting old knowledge. However, if several categories are added to the input graphics, the current model will not be able to deal with it correctly, and the accuracy will drop significantly. Therefore, this paper proposes a kind of incremental method, which adjusts the parameters of the model by identifying the prototype vector and increasing the distance of the vector, so that the model can learn new knowledge without catastrophic forgetting. Experiments show the effectiveness of our proposed method.

cs.CV↗