Dingli Yu
YOU?
Author Swipe
View article: Evaluation of PID Performance at CEPC and Optimization with Combined dN/dx and Time-of-Flight Data
Evaluation of PID Performance at CEPC and Optimization with Combined dN/dx and Time-of-Flight Data Open
This work presents a comprehensive study of charged-hadron particle identification (PID) at the Circular Electron-Positron Collider (CEPC), based on full simulation of hadronic $Z$-pole events. A unified PID strategy is developed by combin…
View article: PaddleOCR 3.0 Technical Report
PaddleOCR 3.0 Technical Report Open
This technical report introduces PaddleOCR 3.0, an Apache-licensed open-source toolkit for OCR and document parsing. To address the growing demand for document understanding in the era of large language models, PaddleOCR 3.0 presents three…
View article: Adaptive Token Boundaries: Integrating Human Chunking Mechanisms into Multimodal LLMs
Adaptive Token Boundaries: Integrating Human Chunking Mechanisms into Multimodal LLMs Open
Recent advancements in multimodal large language models (MLLMs) have demonstrated remarkable capabilities in processing diverse data types, yet significant disparities persist between human cognitive processes and computational approaches …
View article: DeepMath-103K: A Large-Scale, Challenging, Decontaminated, and Verifiable Mathematical Dataset for Advancing Reasoning
DeepMath-103K: A Large-Scale, Challenging, Decontaminated, and Verifiable Mathematical Dataset for Advancing Reasoning Open
Reinforcement learning (RL) with large language models shows promise in complex reasoning. However, its progress is hindered by the lack of large-scale training data that is sufficiently challenging, contamination-free and verifiable. To t…
View article: Weak-to-Strong Generalization Even in Random Feature Networks, Provably
Weak-to-Strong Generalization Even in Random Feature Networks, Provably Open
Weak-to-Strong Generalization (Burns et al., 2024) is the phenomenon whereby a strong student, say GPT-4, learns a task from a weak teacher, say GPT-2, and ends up significantly outperforming the teacher. We show that this phenomenon does …
View article: Optimizing wind turbine blade pitch control via input output differential model free adaptive control
Optimizing wind turbine blade pitch control via input output differential model free adaptive control Open
In the context of wind energy systems, maintaining optimal power output in wind turbines when wind speeds exceed rated values necessitates precise regulation of blade pitch through the pitch control system. However, challenges in accuratel…
View article: AdaGC: Improving Training Stability for Large Language Model Pretraining
AdaGC: Improving Training Stability for Large Language Model Pretraining Open
Large Language Models (LLMs) face increasing loss spikes during scaling, undermining training stability and final performance. While gradient clipping mitigates this issue, traditional global approaches poorly handle parameter-specific gra…
View article: Generalizing from SIMPLE to HARD Visual Reasoning: Can We Mitigate Modality Imbalance in VLMs?
Generalizing from SIMPLE to HARD Visual Reasoning: Can We Mitigate Modality Imbalance in VLMs? Open
Vision Language Models (VLMs) are impressive at visual question answering and image captioning. But they underperform on multi-step visual reasoning -- even compared to LLMs on the same tasks presented in text form -- giving rise to percep…
View article: ADAPTIVE TOKEN BOUNDARIES: INTEGRATING HUMAN CHUNKING MECHANISMS INTO MULTIMODAL LLMS
ADAPTIVE TOKEN BOUNDARIES: INTEGRATING HUMAN CHUNKING MECHANISMS INTO MULTIMODAL LLMS Open
View article: A Geometric Analysis-Based Safety Assessment Framework for Mass Route Decision-Making in Restricted Waters
A Geometric Analysis-Based Safety Assessment Framework for Mass Route Decision-Making in Restricted Waters Open
View article: Deterministic Convergence Analysis for GRU Networks via Smoothing Regularization
Deterministic Convergence Analysis for GRU Networks via Smoothing Regularization Open
View article: Phi-4 Technical Report
Phi-4 Technical Report Open
We present phi-4, a 14-billion parameter language model developed with a training recipe that is centrally focused on data quality. Unlike most language models, where pre-training is based primarily on organic data sources such as web cont…
View article: FlashMask: Efficient and Rich Mask Extension of FlashAttention
FlashMask: Efficient and Rich Mask Extension of FlashAttention Open
The computational and memory demands of vanilla attention scale quadratically with the sequence length $N$, posing significant challenges for processing long sequences in Transformer models. FlashAttention alleviates these challenges by el…
View article: Can Models Learn Skill Composition from Examples?
Can Models Learn Skill Composition from Examples? Open
As large language models (LLMs) become increasingly advanced, their ability to exhibit compositional generalization -- the capacity to combine learned skills in novel ways not encountered during training -- has garnered significant attenti…
View article: ConceptMix: A Compositional Image Generation Benchmark with Controllable Difficulty
ConceptMix: A Compositional Image Generation Benchmark with Controllable Difficulty Open
Compositionality is a critical capability in Text-to-Image (T2I) models, as it reflects their ability to understand and combine multiple concepts from text descriptions. Existing evaluations of compositional capability rely heavily on huma…
View article: AI-Assisted Generation of Difficult Math Questions
AI-Assisted Generation of Difficult Math Questions Open
Current LLM training positions mathematical reasoning as a core capability. With publicly available sources fully tapped, there is unmet demand for diverse and challenging math questions. Relying solely on human experts is both time-consum…
View article: Enhancing the Tracking Performance of Wind Turbine Blade Pitch Angle Control via Model-Free Adaptive Control Algorithm Utilizing Input-Output Differential
Enhancing the Tracking Performance of Wind Turbine Blade Pitch Angle Control via Model-Free Adaptive Control Algorithm Utilizing Input-Output Differential Open
In the context of wind energy systems, maintaining optimal power output in wind turbines when wind speeds exceed rated values necessitates precise regulation of blade pitch through the pitch control system. However, challenges in accuratel…
View article: Keeping LLMs Aligned After Fine-tuning: The Crucial Role of Prompt Templates
Keeping LLMs Aligned After Fine-tuning: The Crucial Role of Prompt Templates Open
Public LLMs such as the Llama 2-Chat underwent alignment training and were considered safe. Recently Qi et al. [2024] reported that even benign fine-tuning on seemingly safe datasets can give rise to unsafe behaviors in the models. The cur…
View article: Mathematical Modeling of Operation Loop Ratio and its Effect in Combat Networks
Mathematical Modeling of Operation Loop Ratio and its Effect in Combat Networks Open
View article: Skill-Mix: a Flexible and Expandable Family of Evaluations for AI models
Skill-Mix: a Flexible and Expandable Family of Evaluations for AI models Open
With LLMs shifting their role from statistical modeling of language to serving as general-purpose AI agents, how should LLM evaluations change? Arguably, a key ability of an AI agent is to flexibly combine, as needed, the basic skills it h…
View article: Tensor Programs VI: Feature Learning in Infinite-Depth Neural Networks
Tensor Programs VI: Feature Learning in Infinite-Depth Neural Networks Open
By classifying infinite-width neural networks and identifying the *optimal* limit, Tensor Programs IV and V demonstrated a universal way, called $μ$P, for *widthwise hyperparameter transfer*, i.e., predicting optimal hyperparameters of wid…
View article: Research on Freezing of Gait Recognition Method Based on Variational Mode Decomposition
Research on Freezing of Gait Recognition Method Based on Variational Mode Decomposition Open
Freezing of Gait (FOG) is the most common and disabling gait disorder in patients with Parkinson’s Disease (PD), which seriously affects the life quality and social function of patients. This paper proposes a FOG recognition method based o…
View article: New Definitions and Evaluations for Saliency Methods: Staying Intrinsic, Complete and Sound
New Definitions and Evaluations for Saliency Methods: Staying Intrinsic, Complete and Sound Open
Saliency methods compute heat maps that highlight portions of an input that were most {\em important} for the label assigned to it by a deep net. Evaluations of saliency methods convert this heat map into a new {\em masked input} by retain…
View article: A Kernel-Based View of Language Model Fine-Tuning
A Kernel-Based View of Language Model Fine-Tuning Open
It has become standard to solve NLP tasks by fine-tuning pre-trained language models (LMs), especially in low-data settings. There is minimal theoretical understanding of empirical success, e.g., why fine-tuning a model with $10^8$ or more…
View article: Program and Organizing Committees
Program and Organizing Committees Open
View article: Table of Contents
Table of Contents Open
View article: Pitch angle control with fault diagnosis and tolerance for wind turbine generation systems
Pitch angle control with fault diagnosis and tolerance for wind turbine generation systems Open
To enhance the reliability of wind turbine generation systems that are generally located in the remote area and subjected to harsh environment, we design the pitch angle control for variable speed wind turbines with the function of fault d…
View article: A New Concept of Fractional Order Cumulant and It-Based Signal Processing in α and/or Gaussian Noise
A New Concept of Fractional Order Cumulant and It-Based Signal Processing in α and/or Gaussian Noise Open
In this article, the concept and definitions of the Fractional Order Moment (FOM) and Fractional Order Cumulant (FOC) are proposed, which is based on the fractional derivative of the fractional order Moment-generating function and the frac…
View article: Adaptive Sliding Mode Control of Lateral Stability of Four Wheel Hub Electric Vehicles
Adaptive Sliding Mode Control of Lateral Stability of Four Wheel Hub Electric Vehicles Open
View article: Distributed Formation Control for Multi-Vehicle Systems with Splitting and Merging Capability
Distributed Formation Control for Multi-Vehicle Systems with Splitting and Merging Capability Open
This letter develops a novel strategy for splitting and merging of agents travelling in formation. The method converts the formation control problem into an optimization problem, which is solved among the agents in a distributed fashion. T…