paper-with-me

Papers

Leveraging Visual Knowledge in Language Tasks: An Empirical Study on Intermediate Pre-training for Cross-modal Knowledge Transfer

2022-03-14 · ACL 2022 5 · Woojeong Jin, Dong-Ho Lee, Chenguang Zhu, Jay Pujara, Xiang Ren

Pre-trained language models are still far from human performance in tasks that need understanding of properties (e.g. appearance, measurable quantity) and affordances of everyday objects in the real world since the text lacks such information due to reporting bias. In this work, we study whether integrating visual knowledge into a language model can fill the gap. We investigate two types of knowledge transfer: (1) text knowledge transfer using image captions that may contain enriched visual knowledge and (2) cross-modal knowledge transfer using both images and captions with vision-language training objectives. On 5 downstream tasks that may need visual knowledge to solve the problem, we perform extensive empirical comparisons over the presented objectives. Our experiments show that visual knowledge transfer can improve performance in both low-resource and fully supervised settings.

📄 PDF Abstract BibTeX arXiv:2203.07519

Code (0)

등록된 구현이 없습니다.

Tasks

Image CaptioningLanguage ModelingLanguage ModellingTransfer Learning

Similar Papers 제목 키워드 기반

Leveraging Visual Knowledge in Language Tasks: An Empirical Study on Intermediate Pre-training for Cross-Modal Knowledge Transfer

2021-11-16 · ACL ARR November 2021 11 · Anonymous

Pre-trained language models are still far from human performance in tasks that need understanding of properties (e.g. appearance, measurable quantity) and affordances of everyday objects in the real world since the text …

Image CaptioningLanguage ModelingLanguage ModellingTransfer Learning

CLIP Models are Few-shot Learners: Empirical Studies on VQA and Visual Entailment

2022-03-14 · ACL 2022 5 · Haoyu Song, Li Dong, Wei-Nan Zhang, Ting Liu 외

CLIP has shown a remarkable zero-shot capability on a wide range of vision tasks. Previously, CLIP is only regarded as a powerful visual encoder. However, after being pre-trained by language supervision from a large amou…

parameter-efficient fine-tuningQuestion AnsweringVisual EntailmentVisual Question Answering+1

Self-Evolving Visual Concept Library using Vision-Language Critics

2025-01-01 · CVPR 2025 1 · Atharva Sehgal, Patrick Yuan, Ziniu Hu, Yisong Yue 외

We study the problem of building a visual concept library for visual recognition. Building effective visual concept libraries is challenging, as manual definition is labor-intensive, while relying solely on LLMs for …

In-Context Learning

Demystifying Language Model Forgetting with Low-rank Example Associations

2024-06-20 · Xisen Jin, Xiang Ren

Large Language models (LLMs) suffer from forgetting of upstream knowledge when fine-tuned. Despite efforts on mitigating forgetting, few have investigated how forgotten upstream examples are dependent on newly learned ta…

Language ModelingLanguage ModellingMatrix Completion

LION: Empowering Multimodal Large Language Model with Dual-Level Visual Knowledge

2024-01-01 · CVPR 2024 1 · Gongwei Chen, Leyang Shen, Rui Shao, Xiang Deng 외

Multimodal Large Language Models (MLLMs) have endowed LLMs with the ability to perceive and understand multi-modal signals. However most of the existing MLLMs mainly adopt vision encoders pretrained on coarsely align…

Language ModelingLanguage ModellingLarge Language ModelMultimodal Large Language Model