# BERT mean pooling

SA-BERT-V1 is a Saudi dialect embedding model fine-tuned based on MARBERTv2, specifically designed for handling Saudi Arabian dialects, providing high-quality sentence embeddings.

This is a Japanese-specific Sentence-BERT model, which uses an improved loss function for training optimization compared to Version 1, achieving a 1.5 to 2 percentage point increase in accuracy.

Finnish sentence BERT model trained based on FinBERT, used for sentence similarity calculation and feature extraction

Representation-based Russian sentence encoder that obtains sentence representations by averaging token embedding vectors