This is a SpanMarker model trained on the conll2003 dataset that can be used for Named Entity Recognition. This SpanMarker model uses roberta-base as the underlying encoder.

Model Details

Model Description

Model Type: SpanMarker
Encoder: roberta-base
Maximum Sequence Length: 256 tokens
Maximum Entity Length: 6 words
Training Dataset: conll2003
Language: en
License: apache-2.0

Model Sources

Repository: SpanMarker on GitHub
Thesis: SpanMarker For Named Entity Recognition

Model Labels

Label	Examples
LOC	"BRUSSELS", "Britain", "Germany"
MISC	"British", "EU-wide", "German"
ORG	"EU", "European Commission", "European Union"
PER	"Werner Zwingmann", "Nikolaus van der Pas", "Peter Blackburn"

Evaluation

Metrics

Label	Precision	Recall	F1
all	0.8944	0.9102	0.9022
LOC	0.9220	0.9215	0.9217
MISC	0.7332	0.7949	0.7628
ORG	0.8764	0.8964	0.8863
PER	0.9605	0.9629	0.9617

Uses

Direct Use for Inference

python

from span_marker import SpanMarkerModel

# Download from the 🤗 Hub
model = SpanMarkerModel.from_pretrained("span_marker_model_id")
# Run inference
entities = model.predict("3. Tristan Hoffman (Netherlands) TVM same time")

Downstream Use

You can finetune this model on your own dataset.

Click to expand

python

from span_marker import SpanMarkerModel, Trainer

# Download from the 🤗 Hub
model = SpanMarkerModel.from_pretrained("span_marker_model_id")

# Specify a Dataset with "tokens" and "ner_tag" columns
dataset = load_dataset("conll2003") # For example CoNLL2003

# Initialize a Trainer using the pretrained model & dataset
trainer = Trainer(
    model=model,
    train_dataset=dataset["train"],
    eval_dataset=dataset["validation"],
)
trainer.train()
trainer.save_model("span_marker_model_id-finetuned")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Sentence length	1	14.5019	113
Entities per sentence	0	1.6736	20

Training Hyperparameters

learning_rate: 1e-05
train_batch_size: 4
eval_batch_size: 4
seed: 42
gradient_accumulation_steps: 2
total_train_batch_size: 8
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 1
mixed_precision_training: Native AMP

Training Results

Epoch	Step	Validation Loss	Validation Precision	Validation Recall	Validation F1	Validation Accuracy
0.2775	500	0.0282	0.9105	0.8355	0.8714	0.9670
0.5549	1000	0.0166	0.9215	0.9205	0.9210	0.9824
0.8324	1500	0.0151	0.9247	0.9346	0.9296	0.9853

Framework Versions

Python: 3.10.12
SpanMarker: 1.5.0
Transformers: 4.41.2
PyTorch: 2.3.0+cu121
Datasets: 2.20.0
Tokenizers: 0.19.1

Citation

BibTeX

@software{Aarsen_SpanMarker,
    author = {Aarsen, Tom},
    license = {Apache-2.0},
    title = {{SpanMarker for Named Entity Recognition}},
    url = {https://github.com/tomaarsen/SpanMarkerNER}
}

bhadauriaupendra062/span-marker-roberta-base-conll03

SpanMarker with roberta-base on conll2003