Multi-lingual BERT Bengali Name Entity Recognition

mBERT-Bengali-NER is a transformer-based Bengali NER model build with bert-base-multilingual-uncased model and Wikiann Datasets.

How to Use

from transformers import AutoTokenizer, AutoModelForTokenClassification
from transformers import pipeline

tokenizer = AutoTokenizer.from_pretrained("sagorsarker/mbert-bengali-ner")
model = AutoModelForTokenClassification.from_pretrained("sagorsarker/mbert-bengali-ner")

nlp = pipeline("ner", model=model, tokenizer=tokenizer, grouped_entities=True)
example = "আমি জাহিদ এবং আমি ঢাকায় বাস করি।"

ner_results = nlp(example)
print(ner_results)

Label and ID Mapping

Label ID Label
0 O
1 B-PER
2 I-PER
3 B-ORG
4 I-ORG
5 B-LOC
6 I-LOC

Training Details

Evaluation Results

Model F1 Precision Recall Accuracy Loss
mBert-Bengali-NER 0.97105 0.96769 0.97443 0.97682 0.12511
Downloads last month
42
Safetensors
Model size
167M params
Tensor type
I64
·
F32
·
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.

Model tree for sagorsarker/mbert-bengali-ner

Finetunes
1 model

Datasets used to train sagorsarker/mbert-bengali-ner