Jerry Yao-Chieh Hu

On Differential Private $\ell_1$, $\ell_2$ and $\ell_p^p$ Distance Queries

Universal Approximation with Softmax Attention

In-Context Algorithm Emulation in Fixed-Weight Transformers

LOBBen-TM: A Benchmark Study of Limit Order Book Prediction with Temporal Modeling

Induction Head Implementation Across Diverse Transformer Weight Constructions

A Theoretical Analysis of Discrete Flow Matching Generative Models

Minimalist Softmax Attention Provably Learns Constrained Boolean Functions

NdLinear: Preserving Multi-Dimensional Structure for Parameter-Efficient Neural Networks

Decoupled Alignment for Robust Plug-and-Play Adaptation

Are Hallucinations Bad Estimations?

Genome-Factory: A Library for Tuning, Deploying, and Interpreting Genomic Models

On Structured State-Space Duality

POLO: Preference-Guided Multi-Turn Reinforcement Learning for Lead Optimization

On Differentially Private String Distances

MathViz-Bench: Evaluating Text-to-Image Models on Visually Solving Math Problems

Pareto-Optimal Energy Alignment for Designing Nature-Like Antibodies

Fast and Low-Cost Genomic Foundation Models via Outlier Removal

Attention Mechanism, Max-Affine Partition, and Universal Approximation

High-Order Flow Matching: Unified Framework and Sharp Statistical Rates

Fundamental Limits of Prompt Tuning Transformers: Universality, Capacity and Efficiency

On Statistical Rates of Conditional Diffusion Transformers: Approximation, Estimation and Minimax Optimality

Nonparametric Modern Hopfield Models

Latent Variable Estimation in Bayesian Black-Litterman Models

In-Context Deep Learning via Transformer Models

In-Context Learning as Conditioned Associative Memory Retrieval

Computational Limits of Low-Rank Adaptation (LoRA) Fine-Tuning for Transformer Models

BOOST: Enhanced Jailbreak of Large Language Model via Slient eos Tokens

SPARQ: Outlier-free SpeechLM with Fast Adaptation and Robust Quantization

Pretrained Transformers are Deep Optimizers: Provable In-Context Learning for Deep Model Training

AlignAb: Pareto-Optimal Energy Alignment for Designing Nature-Like Antibodies

On Differentially Private String Distances

A Benchmark Study For Limit Order Book (LOB) Models and Time Series Forecasting Models on LOB Data

Decoupled Alignment for Robust Plug-and-Play Adaptation