#long-context multimodal learning22/04/2025
NVIDIA's Eagle 2.5: Compact Vision-Language Model Excels in Long-Context Video Tasks Matching GPT-4o
NVIDIA unveiled Eagle 2.5, a compact 8B parameter vision-language model that achieves state-of-the-art performance on long-context video tasks, rivaling much larger models like GPT-4o through innovative training and data strategies.