Nvidia Releases Open-Source NVLM 1.0 to Rival GPT-4
Nvidia unveils NVLM 1.0, a powerful open-source multimodal AI family led by the 72 billion parameter NVLM-D-72B. The model achieves state-of-the-art results to compete directly with proprietary systems from OpenAI and Google.
Nvidia releases a powerful open-source artificial intelligence model that competes directly with proprietary systems from industry leaders like OpenAI and Google. The new NVLM 1.0 family is led by the 72 billion parameter NVLM-D-72B, which demonstrates exceptional performance across both vision and language tasks. By making the model weights publicly available and promising to release the training code, Nvidia breaks from the trend of keeping advanced AI systems closed.
The NVLM-D-72B model shows impressive adaptability in processing complex visual and textual inputs, allowing it to interpret memes, analyze images, and solve mathematical problems step-by-step. Notably, this model improves its performance on text-only tasks after multimodal training, increasing its accuracy by an average of 4.3 points across key text benchmarks. This represents a significant advantage over similar models, which typically see a decline in text performance after adding vision capabilities.
The AI community reacts positively to this open-source initiative, noting that the 72B model performs on par with much larger models in math and coding evaluations. Nvidia's decision to make such a powerful tool openly available accelerates AI research and development across the entire industry. By providing unprecedented access to cutting-edge technology, Nvidia grants researchers and developers the opportunity to build upon a frontier-class system without the restrictions of proprietary licenses.