ByteDance Open-Sources 3B Lance Multimodal Model Trained on 128 A100 GPUs

According to Beating, ByteDance Research open-sourced Lance, a 3-billion-parameter unified multimodal model trained on just 128 A100 GPUs. Lance supports image and video understanding, generation, and editing within a single framework, demonstrating that advanced multimodal capabilities can be achieved with minimal computational resources by using a dual-stream mixture-of-experts architecture and modal-aware rotary position encoding.
Disclaimer: The information on this page may come from third-party sources and is for reference only. It does not represent the views or opinions of Gate and does not constitute any financial, investment, or legal advice. Virtual asset trading involves high risk. Please do not rely solely on the information on this page when making decisions. For details, see the Disclaimer.
Comment
0/400
No comments