Hierarchy-Transformers/HiT-MiniLM-L12-SnomedCT
SNOMED-CT (23) - HiT
Dims:
384
Concepts:
367k
Dataset: Hierarchy-Transformers/SnomedCT
Base Model: all-MiniLM-L12-v2
HiT-MiniLM-L12-SnomedCT is a HiT model trained on SNOMED-CT's concept subsumption hierarchy (TBox).
Hui97/OnT-MiniLM-L12-anatomy
ANATOMY - OnT
Dims:
384
Concepts:
26.7k
Dataset: OnT Zenodo Datasets
Base Model: all-MiniLM-L12-v2
OnT model (ANATOMY), with base model all-MiniLM-L12-v2.
Hui97/OnT-MPNet-anatomy
ANATOMY - OnT
Dims:
768
Concepts:
26.7k
Dataset: OnT Zenodo Datasets
Base Model: all-mpnet-base-v2
OnT model (ANATOMY), with base model all-mpnet-base-v2.
Hui97/OnT-MiniLM-L12-galen
ANATOMY - OnT
Dims:
384
Concepts:
23.1k
Dataset: OnT Zenodo Datasets
Base Model: all-MiniLM-L12-v2
OnT model (GALEN), with base model all-MiniLM-L12-v2.
Hui97/OnT-MPNet-galen
GALEN - OnT
Dims:
768
Concepts:
23.1k
Dataset: OnT Zenodo Datasets
Base Model: all-mpnet-base-v2
OnT model (GALEN), with base model all-mpnet-base-v2.
Hui97/OnT-MiniLM-L12-go
GENE ONTOLOGY - OnT
Dims:
384
Concepts:
51.8k
Dataset: OnT Zenodo Datasets
Base Model: all-MiniLM-L12-v2
OnT model (Gene Ontology), with base model all-MiniLM-L12-v2.
Hui97/OnT-MPNet-go
GENE ONTOLOGY - OnT
Dims:
768
Concepts:
51.8k
Dataset: OnT Zenodo Datasets
Base Model: all-mpnet-base-v2
OnT model (Gene Ontology), with base model all-mpnet-base-v2.
uom-thesis/HiT-mixed-SNOMED-25
SNOMED-CT (25) - HiT
Dims:
384
Concepts:
376k
Dataset: Re-construct (Requires Licensed RF2)
Base Model: all-MiniLM-L12-v2
HiT model trained on SNOMED CT full release (2025).
uom-thesis/OnTr-snomed25-uni
SNOMED-CT (25) - OnT
Dims:
384
Concepts:
376k
Dataset: Re-construct (Requires Licensed RF2)
Base Model: all-MiniLM-L12-v2
OnT model trained on SNOMED CT full release (2025).
uom-thesis/OnTr-m-32
SNOMED-CT (25) - Miniature - OnT
Dims:
384
Concepts:
50k
Dataset: Re-construct (Requires Licensed RF2)
Base Model: all-MiniLM-L12-v2
OnT model trained on SNOMED CT (2025, only: Body Structure, Clinical Finding, Event, Procedure), with a batch size of 32.
uom-thesis/OnTr-minified-64
SNOMED-CT (25) - Miniature - OnT
Dims:
384
Concepts:
50k
Dataset: Re-construct (Requires Licensed RF2)
Base Model: all-MiniLM-L12-v2
OnT model trained on SNOMED CT (2025, only: Body Structure, Clinical Finding, Event, Procedure), with a batch size of 64.
uom-thesis/OnTr-m-128
SNOMED-CT (25) - Miniature - OnT
Dims:
384
Concepts:
50k
Dataset: Re-construct (Requires Licensed RF2)
Base Model: all-MiniLM-L12-v2
OnT model trained on SNOMED CT (2025, only: Body Structure, Clinical Finding, Event, Procedure), with a batch size of 128.
OnT-MiniLM-L12-FULL-EXT-25
SNOMED-CT (25) - Full Onto - Custom OnT
Dims:
384
Concepts:
376k
Dataset: Requires Licensed RF2
Base Model: all-MiniLM-L12-v2
Note that for the full ontology, this appears to be the best performing OnT model (at least, for retrieval tasks). Trained on SNOMED CT (2025, September Release) with a batch size of 64. Trained over ~22k steps. Keywords: cosine annealing, extended training run, downweighted logical loss (0.75). Used in preperation for ACM submission.
HiT-MiniLM-L12-FULL-EXT-25
SNOMED-CT (25) - HiT
Dims:
384
Concepts:
376k
Dataset: Requires Licensed RF2
Base Model: all-MiniLM-L12-v2
HiT model, trained on SNOMED CT (full). Extended training run on H200 GPU instance, marginally better performance.