Large Language Model
models (LMs). Given a fixed budget of tokens, we study how to best select data
A natural way to unlock particular capabilities is to improve this training data.
Learning Mask-aware CLIP Representations for Zero-Shot Segmentation Siyu Jiao 1,2,3, Y unchao Wei 1,2,3, Y aowei Wang
Recently, pre-trained vision-language models have been increasingly used to tackle the challenging zero-shot segmentation task.
Appendices We provide additional experiment details, results and analysis in the supplementary material
T arget: person's face stands out against a swirling, abstract background, their expression one of
Visual Instruction Tuning
Instruction tuning large language models (LLMs) using machine-generated instruction-following data has been shown to improve zero-shot capabilities on new tasks, but the idea is less explored in the multimodal field.
