---
tags:
  - named-entity-recognition
  - clinical-nlp
  - multiclinner
  - multi-head-crf
  - token-classification
language:
  - multilingual
license: apache-2.0
---

# MultiClinNER Multilingual Models

Multilingual clinical NER models trained on all 7 languages (CZ, EN, ES, IT, NL, RO, SV).

## Best Model

- **Model**: `FacebookAI-xlm-roberta-large-C64-H3-E3-Arandom-%0.25-P0.2-42`
- **Branch**: `main`

## Usage

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

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

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

## All Models (7 variants)

| Branch | Model | Best? |
|--------|-------|-------|
| `main` | `FacebookAI-xlm-roberta-large-C64-H3-E3-Arandom-%0.25-P0.2-42` | **Yes** |
| `FacebookAI-xlm-roberta-large-C64-H3-E3-Arandom-pct0.25-P0.2-123` | `FacebookAI-xlm-roberta-large-C64-H3-E3-Arandom-%0.25-P0.2-123` |  |
| `FacebookAI-xlm-roberta-large-C64-H3-E3-Aukn-pct0.1-P0.2-42` | `FacebookAI-xlm-roberta-large-C64-H3-E3-Aukn-%0.1-P0.2-42` |  |
| `FacebookAI-xlm-roberta-large-C64-H3-E3-Aukn-pct0.25-P0.2-123` | `FacebookAI-xlm-roberta-large-C64-H3-E3-Aukn-%0.25-P0.2-123` |  |
| `FacebookAI-xlm-roberta-large-C64-H3-E3-Aukn-pct0.25-P0.2-42` | `FacebookAI-xlm-roberta-large-C64-H3-E3-Aukn-%0.25-P0.2-42` |  |
| `FacebookAI-xlm-roberta-large-C64-H3-E3-Aukn-pct0.25-P0.5-123` | `FacebookAI-xlm-roberta-large-C64-H3-E3-Aukn-%0.25-P0.5-123` |  |
| `FacebookAI-xlm-roberta-large-C64-H3-E3-Aukn-pct0.25-P0.5-42` | `FacebookAI-xlm-roberta-large-C64-H3-E3-Aukn-%0.25-P0.5-42` |  |