Add support for the mixedbread-ai/mxbai-embed-large-v1 which is the best performing model under 1GB on MTEB and also has a ONNX runtime model
Bumped sentence-transformers from 2.2.2 to 2.6.1 as:
a new weightedmean pooling mode has been added since the release of 2.2.1 which is required for mxbai-embed-large-v1
Since 2.3.0 a more efficient form of model loading has been added which should speed up startup times greatly for models that have multiple model types in the repository as only the strictly required files will be downloaded.
mixedbread-ai/mxbai-embed-large-v1which is the best performing model under 1GB on MTEB and also has a ONNX runtime modelsentence-transformersfrom2.2.2to2.6.1as:weightedmeanpooling mode has been added since the release of2.2.1which is required formxbai-embed-large-v12.3.0a more efficient form of model loading has been added which should speed up startup times greatly for models that have multiple model types in the repository as only the strictly required files will be downloaded.