Add NeuML/pubmedbert-base-embeddings (ONNX, biomedical text embeddings)
#22
by krixi-dev - opened
Adds an ONNX version of NeuML/pubmedbert-base-embeddings, a BERT-based sentence embedding model trained on biomedical literature (PubMed).
Why we want it
We run a healthcare social platform for surgeons. Search over medical terminology is the main use case, and a model trained on biomedical text is a natural fit
Details
- Source model: NeuML/pubmedbert-base-embeddings
- Embedding size: 768
- model_type: bert
- Converted with: optimum-cli export onnx --model NeuML/pubmedbert-base-embeddings --task feature-extraction
- The export includes the sentence-embedding output (pooling inside the ONNX graph)
- model_md5 and vocab_md5 in config.json are computed from the files in this PR
- No indexing_prefix or query_prefix is needed
Hi. @Nithish2410 , @harisarang , @jasonbosco , @kishorenc : Could you kindly help review this merge request and approve it. Appreciate your help.