TermMap trained weights (HiLabs AI Quest 2026)

Trained models for TermMap: clinical term + chart context -> ICD-10-CM + SNOMED CT.

file	what
`ce_ICD10CM/`, `ce_SNOMEDCT_US/`	`ncbi/MedCPT-Cross-Encoder` (public domain) fine-tuned per coding system
`ltr_ICD10CM.joblib`, `ltr_SNOMEDCT_US.joblib`	LightGBM LambdaRank rankers
`hierarchy.json`, `coupling.json`, `confidence.json`	calibration: temperatures, ICD category margin, coupling lambda, confidence thresholds

Not included on purpose: the retrieval indexes (contain licensed SNOMED CT descriptions), the training memory, abbreviation table and frequency stats (contain proprietary training terms). termmap prepare rebuilds them from the supplied dictionaries and training CSV. See the repository README for the exact commands.

Downloads last month: -; Downloads are not tracked for this model. How to track

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support