Efficiently deploying visual Transformers on resource-limited PIM systems to minimize inference latency.
