Unifying Multimodal Retrieval via Document Screenshot Embedding
paper : https://arxiv.org/abs/2406.11251
Document Screenshot Embedding (DSE) 제안
제안 배경
https://www.together.ai/blog/multimodal-document-rag-with-llama-3-2-vision-and-colqwen2의 Some Shortcomings of ColPali 부분 참고
