Accelerating Vision-Language Models: BridgeTower on Habana Gaudi2

In three linesHugging Face optimizes BridgeTower, a vision-language model, on Habana Gaudi2 accelerators. The implementation improves inference and training performance for multimodal tasks.

Summary generated by Claude — human-verified