Introduction to Granite Embedding Models

• Huggingface Repository: ibm-granite/granite-embedding-models
• Documentation: Granite Docs
• Granite Community: ibm-granite-community

The Granite Embedding collection delivers innovative sentence-transformer models purpose-built for retrieval-based applications. Featuring a bi-encoder architecture, these models generate high-quality embeddings for textual inputs such as queries, passages, and documents, enabling seamless comparison through cosine similarity. Built using retrieval oriented pretraining, contrastive finetuning, knowledge distillation, and model merging, the Granite Embedding lineup is optimized to ensure strong alignment between query and passage embeddings.

Built on a foundation of carefully curated, permissibly licensed public datasets, the Granite Embedding models set a high standard for performance, maintaining competitive scores not only on academic benchmarks such as BEIR, but also out-perfoming models of the same size on many enterprise use cases. Developed to meet enterprise-grade expectations, they are crafted transparently in accordance with IBM’s AI Ethics principles and offered under the Apache 2.0 license for both research and commercial innovation.

The Granite Embedding lineup includes four different models of varying sizes: - granite-embedding-30m-english: English only model that produces embedding vectors of size 384. - granite-embedding-125m-english: English only model that produces embedding vectors of size 768. - granite-embedding-107m-multilingual: Multilingual model that produces embedding vectors of size 384. - granite-embedding-278m-multilingual: Multilingual model that produces embedding vectors of size 768.

Accordingly, these options provide a range of models with different compute requirements to choose from, with appropriate trade-offs with their performance on downstream tasks.