Nvidia Unveils Open-Source NVLM 1.0, Achieves State-of-the-Art Vision-Language Results

Sep 18, 2024, 02:33 AM

Nvidia has introduced NVLM 1.0, a new family of open-source frontier-class multimodal large language models (LLMs) that achieve state-of-the-art results on vision-language tasks. These models rival leading proprietary models such as GPT-4o and open-access models like InternVL 2. NVLM 1.0 also shows improved performance on text-only tasks following multimodal training. The code and model weights for NVLM 1.0 are now open-source. The technical report accompanying the release provides insights into architectural design, data curation, and other aspects of building cutting-edge multimodal LLMs.