large language models 1self-improving models 1shortcut connections 1training-free finetuning 1vision-language models 1
From the 1 of 3 linked papers with an AI index.
3 papers
cs.LG2026
Self-Improving is Often Sudden: Enlightenment-style Finetuning for Large-Scale Models
Jing-Xiao Liao, Tianwei Zhang, Yu-Hao Jiang +3
The paper proposes Enlightenment, a training-free post‑tuning method that adds shortcut connections to key layers of large language and vision‑language models, enabling sudden perf…
cs.CV2026
When Can We Trust Deep Neural Networks? Towards Reliable Industrial Deployment with an Interpretability Guide
Hang-Cheng Dong, Yuhao Jiang, Yibo Jiao +5
The deployment of AI systems in safety-critical domains, such as industrial defect inspection, autonomous driving, and medical diagnosis, is severely hampered by their lack of reli…
cs.LG2026
Quotient Geometry, Effective Curvature, and Implicit Bias in Simple Shallow Neural Networks
Hang-Cheng Dong, Pengcheng Cheng
Overparameterized shallow neural networks admit substantial parameter redundancy: distinct parameter vectors may represent the same predictor due to hidden-unit permutations, resca…