Back to AI Lab
Robustness
Research papers, repositories, and articles about robustness
Showing 2 of 2 items
Neuro-RIT: Neuron-Guided Instruction Tuning for Robust Retrieval-Augmented Language Model
Introduces Neuro-RIT, which looks at individual neurons while customizing language models for retrieval-heavy tasks. The aim is steadier answers when retrieved documents shift or are noisy.
Jaemin Kim, Jae O Lee
Not All Relations Rotate Alike: Transformation-Aware Decoupling for Viewpoint-Robust 3D Scene Graph Generation
Separates 3D scene relations into ones that should rotate with the camera and ones that should stay fixed. This decoupling makes scene graphs much more stable under viewpoint changes. If you're building embodied agents or 3D perception, this is a practical robustness trick. ([arxiv.org](https://arxiv.org/list/cs.CV/new))
Jingjun Sun, Chaowei Wang