IEETA/MultiClinNER-NL

MultiClinNER NL Models

Clinical NER models for NL, trained with Multi-Head CRF architecture.

Best Model

• Model: CLTL-C64-H3-E60-Arandom-%0.25-P0.2-42
• Best F1: 0.6898
• Branch: main

Usage

# Load the best model (main branch)
from transformers import AutoTokenizer, AutoModelForTokenClassification

model = AutoModelForTokenClassification.from_pretrained("IEETA/MultiClinNER-NL")
tokenizer = AutoTokenizer.from_pretrained("IEETA/MultiClinNER-NL")

# Load a specific model variant
model = AutoModelForTokenClassification.from_pretrained("IEETA/MultiClinNER-NL", revision="BRANCH_NAME")

All Models (18 variants)

Branch	Model	Best?
main	CLTL-C64-H3-E60-Arandom-%0.25-P0.2-42	Yes
CLTL-C64-H3-E60-Anone-pct0.2-P0.5-123	CLTL-C64-H3-E60-Anone-%0.2-P0.5-123
CLTL-C64-H3-E60-Anone-pct0.2-P0.5-42	CLTL-C64-H3-E60-Anone-%0.2-P0.5-42
CLTL-C64-H3-E60-Anone-pct0.2-P0.5-456	CLTL-C64-H3-E60-Anone-%0.2-P0.5-456
CLTL-C64-H3-E60-Anone-pct0.2-P0.5-999	CLTL-C64-H3-E60-Anone-%0.2-P0.5-999
CLTL-C64-H3-E60-Arandom-pct0.1-P0.5-123	CLTL-C64-H3-E60-Arandom-%0.1-P0.5-123
CLTL-C64-H3-E60-Arandom-pct0.1-P0.5-42	CLTL-C64-H3-E60-Arandom-%0.1-P0.5-42
CLTL-C64-H3-E60-Arandom-pct0.1-P0.5-456	CLTL-C64-H3-E60-Arandom-%0.1-P0.5-456
CLTL-C64-H3-E60-Arandom-pct0.25-P0.2-123	CLTL-C64-H3-E60-Arandom-%0.25-P0.2-123
CLTL-C64-H3-E60-Arandom-pct0.25-P0.2-999	CLTL-C64-H3-E60-Arandom-%0.25-P0.2-999
CLTL-C64-H3-E60-Aukn-pct0.1-P0.2-123	CLTL-C64-H3-E60-Aukn-%0.1-P0.2-123
CLTL-C64-H3-E60-Aukn-pct0.1-P0.2-42	CLTL-C64-H3-E60-Aukn-%0.1-P0.2-42
CLTL-C64-H3-E60-Aukn-pct0.1-P0.2-456	CLTL-C64-H3-E60-Aukn-%0.1-P0.2-456
CLTL-C64-H3-E60-Aukn-pct0.1-P0.2-999	CLTL-C64-H3-E60-Aukn-%0.1-P0.2-999
CLTL-C64-H3-E60-Aukn-pct0.1-P0.5-123	CLTL-C64-H3-E60-Aukn-%0.1-P0.5-123
CLTL-C64-H3-E60-Aukn-pct0.1-P0.5-42	CLTL-C64-H3-E60-Aukn-%0.1-P0.5-42
CLTL-C64-H3-E60-Aukn-pct0.1-P0.5-456	CLTL-C64-H3-E60-Aukn-%0.1-P0.5-456
CLTL-C64-H3-E60-Aukn-pct0.1-P0.5-999	CLTL-C64-H3-E60-Aukn-%0.1-P0.5-999