A RoBERTa-based sentence embedding model fine-tuned on the STS Benchmark for semantic textual similarity.