How AI describes GTE-large

Alibaba-NLP's gte-large-en-v1.5 is part of the gte-v1.5 series, providing long-context text embeddings (up to 8192 tokens) for multilingual text retrieval and reranking. It uses a Transformer++ encoder backbone (BERT + RoPE + GLU) and achieves state-of-the-art MTEB scores within its model size, with the gte-Qwen1.5-7B-instruct variant for multilingual instruction-tuned embedding. The model is available on Hugging Face for easy integration with transformers, sentence-transformers, and transformers.js to compute embeddings and similarity for semantic search and retrieval tasks.