Preprint2024
NVLM
Unknown
NVLM introduces a family of multimodal LLMs with a hybrid architecture and 1-D tile-tagging for dynamic high-resolution images, improving text-only and multimodal performance.
0Sep 1, 2024Large Language ModelsAttention Mechanisms
arXiv