Models

This release contains all four models, including tokenizers, for named entity extraction. See below the average performance, measured by entity-level f1-score.

ICD/NDC Embedding Representations

The embedding-representations for the ICD and NDC codes are appended to this release as numpy files. To use them, put them into the data folder before executing the pipeline.