CardioEmbed · benchmark
Domain specialization, measured.
How a LoRA-adapted clinical-cardiology embedding compares against five baseline models on the published benchmark. Real numbers from arXiv:2511.10930. Model weights on Hugging Face.
Separation score · higher is better
The benchmark.
Separation score measures how well a model differentiates clinical cardiology concepts that look textually similar but mean different things (e.g. aortic stenosis vs aortic regurgitation). CardioEmbed's LoRA adaptation gives it a 3.0× advantage over the strongest biomedical baseline.
Flagship LoRA-adapted embedding for clinical cardiology. Best-in-class separation, 99.6% classification accuracy.
Efficient 2B-parameter baseline. Strong balance of speed and accuracy for production workloads.
Latest Qwen3 architecture. Multilingual support for international deployments.
Ultra-lightweight baseline. Fits on edge and resource-constrained environments.
Industry-standard E5 embedding model. Common baseline for semantic search.
Pre-trained on PubMed citations. Strong general biomedical NLP baseline.
All metrics
What the numbers mean.
| Model | Separation | Improvement | Dim | Accuracy |
|---|---|---|---|---|
| CardioEmbed · Qwen3-8B | 0.510 | +1452% | 3584 | 99.6% |
| Gemma-2-2B | 0.455 | +700% | 2304 | — |
| Qwen3-4B | 0.446 | +1184% | 1792 | — |
| Qwen2.5-0.5B | 0.327 | +1215% | 896 | — |
| E5-large-v2 | 0.284 | +787% | 1024 | — |
| BioLinkBERT-Large | 0.168 | +438% | 1024 | — |
Next
Run it yourself.
- ·Download the model from Hugging Face →
- ·Read the paper on arXiv →
- ·See the full LoRA product lineup →
- ·Back to all demos →
Live inference endpoint (deepneuro.ai/lora-api) coming once the RunPod deployment lands. Until then, the model weights on Hugging Face are the canonical way to run it yourself.