Collections

Highly efficient Mixture-of-Experts (MoE) large language model for advanced reasoning and text generation.

Microsoft's harrier is a decoder-only, multilingual text embedding model family using last-token pooling and L2 normalization to generate Embeddings