Embeddings and rerankers · BAAI
BGE-M3
Dense, sparse and multi-vector retrieval in one model; 8K context.
Datasheet
Maker | BAAI |
|---|---|
Parameters | —Not yet measuredHF API lists no safetensors total for this repo Source |
Licence | MITVerified |
Access | Public download |
Task (HF pipeline) | sentence-similarity |
First published on HF | 2024-01-27 |
Last updated on HF | 2024-07-03 |
Benchmark
59.56
MMTEB (multilingual) mean task score; rerankers: MTEB-R · #4 of 4 scored in this category
Row in the Qwen3-Embedding card table (taken from the MTEB leaderboard, 2025-05-24). Source
The description summarises the model card at https://huggingface.co/BAAI/bge-m3. Parameters and licence are read from the Hugging Face API; the weights-in-memory line is our arithmetic, not a measurement.