Hi thanks for your interest in my article! As to your question, I think there are no out-of-the box Gujarati models trained on semantic similarity, which means you might have to find a pretrained Gujarati model and fine-tune it yourself on a Gujarati semantic similarity dataset. You can find some Gujarati pretrained models here: https://huggingface.co/models?filter=gu

Alternatively, you can also try translating Gujarati text into English and then follow this article to calculate the semantic similarity. This might be an easier approach, considering you can use the pretrained Gujarati-English translation model provided by HelsinkiNLP here: https://huggingface.co/Helsinki-NLP/opus-mt-mul-en