Article

When Lower Reconstruction Loss Hurts: Distributionally Robust Refinement for Low-Bit LLM Quantization
Weight-only post-training quantization (PTQ) relies heavily on reconstruction loss minimization to preserve model quality at low …
PDF Cite DOI
Discrete Wasserstein Flows for One-Step Generative Modeling
We introduce a new framework for one-step generative modelling on finite state spaces. To extend drifting beyond continuous domains, we …
PDF Cite DOI
Sequence-Informed Geometric Evaluation of RNA 3D Structures
Computational RNA structure pipelines generate many candidate conformations for the same sequence. Reliable evaluation therefore …
PDF Cite DOI
SR-OPSD: Self-Referenced On-Policy Self-Distillation
On-policy self-distillation (OPSD) converts feedback into dense token-level supervision on student-generated trajectories, …
PDF Cite DOI
Definitions and Metrics of Interdisciplinarity: A PRISMA-ScR Scoping Review Protocol
This project is a PRISMA-ScR scoping review protocol titled “Definitions and Metrics of Interdisciplinarity,” which aims to …
PDF Cite DOI
Generative Modeling on Metric Graphs via Neural Optimal Transport
We introduce, to our knowledge, the first deep generative modeling framework for probability distributions continuously supported on …
PDF Cite DOI
Federated analysis of incubation period distributions using individual-level observed data and heterogeneous summary statistics
Abstract The incubation period, the interval between pathogen exposure and symptom onset, is a critical epidemiological parameter for …
PDF Cite DOI
PlayClass: Automated Play Behaviour Classification in Poultry
Automated monitoring of animal welfare has largely targeted negative indicators, leaving positive welfare behaviours such as play …
PDF Cite DOI
Interdomain Attention: Beyond Token-Level Key-Value Memory
Transformers and deep state space models (SSMs) sit at opposite ends of a basic design choice: attention routes each query through a …
PDF Cite DOI
Saliency-Aware Regularized Quantization Calibration for Large Language Models
Post-training quantization (PTQ) is an effective approach for deploying large language models (LLMs) under memory and latency …
PDF Cite DOI