metadata
license: mit
language:
- en
tags:
- medical
pretty_name: ICD10CM embeddings with intfloat/e5_small_v2
size_categories:
- 10K<n<100K
configs:
- config_name: default
data_files:
- split: train
path: data/train-*
dataset_info:
features:
- name: order
dtype: int64
- name: code
dtype: string
- name: valid
dtype: bool
- name: short_desc
dtype: string
- name: long_desc
dtype: string
- name: embedding
sequence: float64
- name: level
dtype: int64
splits:
- name: train
num_bytes: 315274220
num_examples: 97584
download_size: 3513277
dataset_size: 315274220
Embeddings of icd10cm-order-2025.txt
sourced from cdc.gov,
using intfloat/e5_small_v2
, a solid performer for its size on ICD10CM, see https://arxiv.org/abs/2401.01943.
Input labels are prefixed with query:
per model guidance.
- code: The ICD10-CM code
- level: the level of the code; 0 for A01, 1 for A01.0, 2 for A01.00, etc.
- parent_code: parent ICD10-CM code; "None" for level-0 codes.
- valid: whether the code is valid.
- embedded_text: the embedded text (the label prefixed with
query:
) - embedding: comma-separated vector