From the model card
Fields below are copied from the tags and counters on the HuggingFace repository ncbi/MedCPT-Cross-Encoder at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.
- Publisher (HF namespace)
- ncbi
- Pipeline tag
- text-classification
- Library
- Transformers
- Framework tags
- PyTorch
- License tag
other— read the license file in the repo before relying on it- Downloads (HF counter at last fetch)
- 496,643
- Likes (HF counter at last fetch)
- 32
- Model card
- https://huggingface.co/ncbi/MedCPT-Cross-Encoder
Use cases
- Reranking retrieved PubMed abstracts for clinical question answering
- Scoring relevance of medical literature to patient case summaries
- Building clinical decision support retrieval pipelines
- Evaluating biomedical RAG system retrieval quality
- Re-ordering candidate answers in medical QA benchmarks
Pros
- Trained on PubMed clinical query pairs; strong domain-specific relevance scoring
- Text-embeddings-inference compatible for efficient batched reranking
- 30 likes with documented use in biomedical IR research
- Complements the MedCPT bi-encoder for a full two-stage retrieval system
Cons
- Cross-encoder inference is quadratic in candidate count; not suitable for first-stage retrieval
- PubMed-centric training may not generalise to clinical notes or EHR text
- No explicit license declared; verify before clinical deployment
- English only; no multilingual medical retrieval support
Tags
transformerspytorchberttext-classificationlicense:othertext-embeddings-inferenceendpoints_compatibleregion:usdeploy:azure