We are pleased to announce the publication of our paper "Semantic similarity prediction is better than other semantic similarity measures" in the Transactions on Machine Learning and Research (TMLR). The research challenges conventional methods of measuring semantic similarity in natural language texts. We introduce the STSScore approach, utilizing a fine-tuned model for the Semantic Textual Similarity Benchmark tasks from the GLUE benchmark. The findings demonstrate that directly predicting semantic similarity through this approach surpasses traditional methods such as BLEU and embeddings like BERTScore and S-BERT.