Loading the SOTA2 catalog…
QATMA: Quantization-Aware Training with Multimodal Alignment for Open-Vocabulary Object Detection · SOTA2 Research