bert base multilingual cased ner hrl

Introduction

The bert-base-multilingual-cased-ner-hrl is a Named Entity Recognition (NER) model designed to identify entities such as locations, organizations, and persons across ten high-resourced languages. It is a fine-tuned version of the multilingual BERT (mBERT) base model.

Architecture

The model is based on the bert-base-multilingual-cased architecture, which supports multiple languages including Arabic, German, English, Spanish, French, Italian, Latvian, Dutch, Portuguese, and Chinese. It uses a token classification approach to recognize entities in text.

Training

The model was fine-tuned using datasets specific to each language:

• Arabic: ANERcorp
• German: CoNLL 2003
• English: CoNLL 2003
• Spanish: CoNLL 2002
• French: Europeana Newspapers
• Italian: Italian I-CAB
• Latvian: Latvian NER
• Dutch: CoNLL 2002
• Portuguese: Paramopama + Second Harem
• Chinese: MSRA

It was trained on an NVIDIA V100 GPU using Hugging Face's recommended hyperparameters.

Guide: Running Locally

To run the model locally, follow these steps:

1. Install the Transformers library:

   pip install transformers
   

2. Load the tokenizer and model:

   from transformers import AutoTokenizer, AutoModelForTokenClassification, pipeline
   
   tokenizer = AutoTokenizer.from_pretrained("Davlan/bert-base-multilingual-cased-ner-hrl")
   model = AutoModelForTokenClassification.from_pretrained("Davlan/bert-base-multilingual-cased-ner-hrl")
   nlp = pipeline("ner", model=model, tokenizer=tokenizer)
   

3. Perform Named Entity Recognition:

   example = "Nader Jokhadar had given Syria the lead with a well-struck header in the seventh minute."
   ner_results = nlp(example)
   print(ner_results)
   

For improved performance, consider using cloud GPUs such as those available on AWS or Google Cloud.

License

The model is released under the Academic Free License v3.0 (AFL-3.0).