Sentence Similarity
sentence-transformers
ONNX
Safetensors
Chinese
English
modernbert
embeddings
medical
clinical-nlp
healthcare
biomedical
fhir
traditional-chinese
taiwan
rag
text-embeddings-inference
Instructions to use weemed/IlhaEmbed with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use weemed/IlhaEmbed with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("weemed/IlhaEmbed") sentences = [ "帶狀皰疹", "皮蛇", "高血壓用藥", "114年成健" ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [4, 4] - Notebooks
- Google Colab
- Kaggle
Commit History
-
release: update IlhaEmbed with TAIDE-LX-7B 3-epoch knowledge distillation & INT8 ONNX (36.78MB)
Verified signature823a8a4
-
release: update IlhaEmbed v3.1 Omni weights, INT8 ONNX (36.8MB) & benchmark evidence
Verified signature1556437
-
Release IlhaEmbed-97M Edge v4: Multi-Head Attention Pooling & RxNorm 85.99%
Verified signaturePromote candidate trial-0004 (unbiased source-conditioned attention-pooling unfrozen ModernBERT). - RxNorm Prescription Top-1: 85.99% (Baseline: 58.89%, +27.09% absolute gain, p < 10^-5) - RxNorm Prescription Top-5: 94.97% (Baseline: 74.18%, +20.79% absolute gain) - DrugBank Terminology Top-1: 63.47% (Baseline: 52.10%, +11.38% absolute gain, p < 10^-5) - Clinical Shorthand Top-1: 92.56% (Top-5: 97.52%) - Taigi Medical Semantics Top-1: 93.62% (Top-5: 96.45%) - Edge CPU Inference: 3.97 ms / text (254 texts/s throughput on CPU) - INT8 Dynamic Quantization Parity: 0.9955 mean cosine (min 0.9897) - Ships full ModernBERT backbone, 4-head Attention Pooling, INT8 ONNX, and compact 37MB ONNX.05d0db6