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
-
docs: update README with TAIDE 3-epoch distillation benchmark & honest failure case breakdown
Verified signature1668c2f -
release: update IlhaEmbed with TAIDE-LX-7B 3-epoch knowledge distillation & INT8 ONNX (36.78MB)
Verified signature823a8a4
-
docs: update MTEB benchmark data with official RTX 4080 rerun scores
Verified signaturead11391 -
release: bump version to v3.1 across English and Traditional Chinese documentation
Verified signature8f72b8f -
fix: strictly bold winning scores across MTEB benchmark tables
Verified signature837ab89 -
revert: restore complete original bilingual (English & Traditional Chinese) Model Card
Verified signatureba57efd -
docs: remove defensive copy and keep clean benchmark presentation
Verified signatureda5e196 -
docs: restore full comparative benchmark with competitors and transparent analysis
Verified signatureaffd64c -
docs: update verified official MTEB and clinical benchmark scores
Verified signature9b8396e -
docs: refine terminology to 平均準確率 (Macro Avg)
Verified signature391355c -
docs: use standard Taiwan engineering term Macro 平均
Verified signatureca2f4c1 -
docs: use authentic Taiwan terminology 巨觀平均 (Macro-average)
Verified signature0492052 -
release: update IlhaEmbed v3.1 Omni weights, INT8 ONNX (36.8MB) & benchmark evidence
Verified signature1556437
-
feat(brand): add 17th C. Ilha Formosa horizontal nautical line-art logo to header
Verified signature7c5463b -
feat(brand): add 17th C. Ilha Formosa horizontal nautical line-art logo asset
Verified signature05e01a8 -
fix(style): remove block img from headings to ensure clean typography and proper heading alignment
Verified signature92e2fa4 -
style(readme): replace informal emojis with clean institutional vector SVG icons
Verified signature2ae08aa -
feat(assets): add vector SVG icon scale.svg
Verified signaturebb11ef3 -
feat(assets): add vector SVG icon shield.svg
Verified signature5dd3498 -
feat(assets): add vector SVG icon terminal.svg
Verified signatured0110b3 -
feat(assets): add vector SVG icon clipboard.svg
Verified signature94934d3 -
feat(assets): add vector SVG icon flask.svg
Verified signaturecb4f100 -
feat(assets): add vector SVG icon chart.svg
Verified signature7f24ddd -
feat(assets): add vector SVG icon sparkles.svg
Verified signature4a53559 -
docs: add official MTEB benchmark comparison and 20-note clinical evaluation report
Verified signature9b513a9 -
docs: fix CJK markdown bold rendering and optimize community tags with interactive widget
Verified signature5f576de -
docs: restore canonical English-first Chinese-second README and replace model_int8.onnx with 37MB compact edge model
Verified signaturecf74485 -
clean: remove internal research artifacts and streamline model card
Verified signatureKeep only user-facing model weights, tokenizer configs, ONNX runtime files, and a clean README.2487113 -
docs: restore clean 311M flagship benchmarks and isolate 97M edge retraining table
Verified signatureb31f0f8 -
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 -
Release IlhaEmbed-97M Edge: 100% (44/44) FHIR prototype routing & INT8 ONNX
Verified signaturefd065f2 -
docs: update model card benchmark to 100% FHIR routing for Edge 97M
Verified signaturee0f135a -
docs: sync model card from WeeMed/ilhaembed
Verified signature3dbfe69
-
docs: sync model card from WeeMed/ilhaembed
Verified signature0d043b0
-
Correct v2.1.0 status: automatic classification is not qualified
Verified signature99d4f90 -
Release IlhaEmbed v2.1.0 with corrected category regression evaluation
Verified signaturePublish the exact validated concept-adapted FP32 and INT8 artifacts. Owner-confirmed classification counts correct, wrong-category and unclassified outcomes separately; unresolved residue is not rejection gold. Record regression results, numerical and CPU evidence, source lineage and limitations.91628ef
-
docs: remove internal legacy CODER-TW references from SOURCES.md
Verified signaturec591f85 -
docs: update README.md Model Card with v2.0 clinical specs and benchmark tables
Verified signaturea2dd71d -
release: update IlhaEmbed with expanded clinical lab & procedural contrastive tuning
Verified signatureee78911 -
release: update IlhaEmbed with expanded clinical lab & procedural contrastive tuning
Verified signaturef921016 -
release: update IlhaEmbed with expanded clinical lab & procedural contrastive tuning
Verified signaturecda8e25 -
release: update IlhaEmbed with expanded clinical lab & procedural contrastive tuning
Verified signature12da14e -
release: update IlhaEmbed with expanded clinical lab & procedural contrastive tuning
Verified signature461af5a
-
Release corrected source-isolated IlhaEmbed v2
Verified signature1dae156
-
docs: update Hugging Face main README model card for IlhaEmbed v2.0
Verified signature7bca3df -
docs: update Hugging Face main README model card for IlhaEmbed v2.0
Verified signatureadbc0a5 -
feat: update model_int8.onnx (IlhaEmbed v2 NAER retrained INT8 ONNX)
Verified signature50eba12 -
feat: upload IlhaEmbed v2 model weights and tokenizer (NAER 13-set retrained)
Verified signaturea922e01
-
docs: add Context-Conditioned usage and MODA Data Provenance to Traditional Chinese section
Verified signature1fd5af6 -
docs: fix strikethrough markdown formatting caused by unescaped tildes
Verified signaturea815c86