BGE
Embedding model family that ranks well on retrieval benchmarks.
Visão geral
BGE is a family of embedding models trained specifically for retrieval, available in sizes small enough to run locally. It solves the problem that a general-purpose embedding model is not optimised for the asymmetric comparison a search query and a document require. It exists because retrieval quality is set at the embedding step, so a model trained for it improves everything downstream without changing anything else.
Grafo de tecnologias
Como BGE se liga ao resto do stack.
Ferramentas
Software que implementa ou se integra com BGE.
Benchmarks
LLM Frameworks
Alternativas
-
Cohere Embed
model
Uma alternativa a BGE · source
-
E5
model
Uma alternativa a BGE · source
-
GTE
model
Uma alternativa a BGE · source
-
Instructor
model
Uma alternativa a BGE · source
-
Jina Embeddings
model
Uma alternativa a BGE · source
-
Nomic Embed
model
Uma alternativa a BGE · source
-
OpenAI Text Embedding 3
model
Uma alternativa a BGE · source
Comparações
Comparações detalhadas com as alternativas acima.