UniRaG: Unification, Retrieval, and Generation for Multimodal Question Answering With Pre-Trained Language Models

Keyphrases

• Multimodal Question Answering 100%
• Question Generation 50%
• Text-based Approach 33%
• Context Retrieval 33%
• Negative Context 16%
• Sequence Classification 16%
• Distractor 16%
• Top-k Retrieval 16%
• Text Representation 16%
• Cross-encoder 16%
• Three-stage Framework 16%
• Contextual Diversity 16%
• Image Captioning 16%
• Retrieval Strategies 16%
• Multimodal Comprehension 16%
• Flan 16%
• Diversity Training 16%

Earth and Planetary Sciences

• Knowledge Representation 100%
• Coder 100%
• Natural Language Processing 100%
• State of the Art 50%
• Computer Vision 50%

Computer Science

• Sequence Classification 33%
• Image Captioning 33%
• Transformation Process 33%
• Diversity Training 33%
• Textual Representation 33%