There is a scarcity of evidence and benchmarks indicating the performance of sentence embedding transformers on Arabic language text, especially for smaller models that can be trained on a single ...