---
license: mit
language:
- en
- fr
- de
- it
- es
- pt
- pl
- nl
- ru
pipeline_tag: token-classification
inference: false
tags:
- token-classification
- entity-recognition
- foundation-model
- feature-extraction
- mBERT
- Multilingual Bert
- BERT
- generic
---

# SOTA Entity Recognition Multilingual Foundation Model by NuMind 🔥

 This model provides the best embedding for the Entity Recognition task and supports 9+ languages.

**Checkout other models by NuMind:**
* SOTA Entity Recognition Foundation Model in English: [link](https://huggingface.co/numind/entity-recognition-general-sota-v1)
* SOTA Sentiment Analysis Foundation Model: [English](https://huggingface.co/numind/generic-sentiment-v1), [Multilingual](https://huggingface.co/numind/generic-sentiment-multi-v1)

## About

[Multilingual BERT](https://huggingface.co/bert-base-multilingual-cased) finetunned on an artificially annotated multilingual subset of [Oscar dataset](https://huggingface.co/datasets/oscar-corpus/OSCAR-2201). This model provides domain & language independent embedding for Entity Recognition Task. We fine-tunned it only on 9 languages but the model can generalize over other languages that are supported by the Multilingual BERT.

**Metrics:**

Read more about evaluation protocol & datasets in our [blog post](https://www.numind.ai/blog/a-foundation-model-for-entity-recognition)

| Model | F1 macro |
|----------|----------|
|   bert-base-multilingual-cased  |  0.5206   |
|   ours  |   0.5892  |
|   ours + two emb  |   0.6231  |


## Usage

Embeddings can be used out of the box or fine-tuned on specific datasets. 

Get embeddings:

```python
import torch
import transformers


model = transformers.AutoModel.from_pretrained(
    'numind/entity-recognition-multilingual-general-sota-v1',
    output_hidden_states=True,
)
tokenizer = transformers.AutoTokenizer.from_pretrained(
    'numind/entity-recognition-multilingual-general-sota-v1',
)

text = [
    "NuMind is an AI company based in Paris and USA.",
    "NuMind est une entreprise d'IA basée à Paris et aux États-Unis.",
    "See other models from us on https://huggingface.co/numind"
]
encoded_input = tokenizer(
    text,
    return_tensors='pt',
    padding=True,
    truncation=True
)
output = model(**encoded_input)

# two emb trick: for better quality
emb = torch.cat(
    (output.hidden_states[-1], output.hidden_states[-7]),
    dim=2
)

# single emb: for better speed
# emb = output.hidden_states[-1]
```