This is a SpanMarker model trained on the FewNERD dataset that can be used for Named Entity Recognition. This SpanMarker model uses BAAI/bge-base-en-v1.5 as the underlying encoder.

Model Details

Model Description

Model Type: SpanMarker
Encoder: BAAI/bge-base-en-v1.5
Maximum Sequence Length: 256 tokens
Maximum Entity Length: 8 words
Training Dataset: FewNERD
Language: en
License: cc-by-sa-4.0

Model Sources

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

Model Labels

Label	Examples
art-broadcastprogram	"Corazones", "The Gale Storm Show : Oh , Susanna", "Street Cents"
art-film	"L'Atlantide", "Bosch", "Shawshank Redemption"
art-music	"Atkinson , Danko and Ford ( with Brockie and Hilton )", "Hollywood Studio Symphony", "Champion Lover"
art-other	"Venus de Milo", "The Today Show", "Aphrodite of Milos"
art-painting	"Cofiwch Dryweryn", "Touit", "Production/Reproduction"
art-writtenart	"Time", "The Seven Year Itch", "Imelda de ' Lambertazzi"
building-airport	"Newark Liberty International Airport", "Luton Airport", "Sheremetyevo International Airport"
building-hospital	"Hokkaido University Hospital", "Memorial Sloan-Kettering Cancer Center", "Yeungnam University Hospital"
building-hotel	"Radisson Blu Sea Plaza Hotel", "Flamingo Hotel", "The Standard Hotel"
building-library	"Bayerische Staatsbibliothek", "British Library", "Berlin State Library"
building-other	"Communiplex", "Alpha Recording Studios", "Henry Ford Museum"
building-restaurant	"Carnegie Deli", "Fatburger", "Trumbull"
building-sportsfacility	"Boston Garden", "Glenn Warner Soccer Facility", "Sports Center"
building-theater	"Pittsburgh Civic Light Opera", "National Paris Opera", "Sanders Theatre"
event-attack/battle/war/militaryconflict	"Jurist", "Vietnam War", "Easter Offensive"
event-disaster	"1693 Sicily earthquake", "the 1912 North Mount Lyell Disaster", "1990s North Korean famine"
event-election	"Elections to the European Parliament", "March 1898 elections", "1982 Mitcham and Morden by-election"
event-other	"Eastwood Scoring Stage", "Masaryk Democratic Movement", "Union for a Popular Movement"
event-protest	"French Revolution", "Iranian Constitutional Revolution", "Russian Revolution"
event-sportsevent	"Stanley Cup", "National Champions", "World Cup"
location-GPE	"Croatian", "the Republic of Croatia", "Mediterranean Basin"
location-bodiesofwater	"Norfolk coast", "Atatürk Dam Lake", "Arthur Kill"
location-island	"Staten Island", "Laccadives", "new Samsat district"
location-mountain	"Ruweisat Ridge", "Salamander Glacier", "Miteirya Ridge"
location-other	"Victoria line", "Northern City Line", "Cartuther"
location-park	"Shenandoah National Park", "Gramercy Park", "Painted Desert Community Complex Historic District"
location-road/railway/highway/transit	"NJT", "Friern Barnet Road", "Newark-Elizabeth Rail Link"
organization-company	"Texas Chicken", "Dixy Chicken", "Church 's Chicken"
organization-education	"Barnard College", "MIT", "Belfast Royal Academy and the Ulster College of Physical Education"
organization-government/governmentagency	"Diet", "Congregazione dei Nobili", "Supreme Court"
organization-media/newspaper	"Clash", "TimeOut Melbourne", "Al Jazeera"
organization-other	"Defence Sector C", "IAEA", "4th Army"
organization-politicalparty	"Al Wafa ' Islamic", "Kenseitō", "Shimpotō"
organization-religion	"Christian", "Jewish", "UPCUSA"
organization-showorganization	"Lizzy", "Mr. Mister", "Bochumer Symphoniker"
organization-sportsleague	"First Division", "China League One", "NHL"
organization-sportsteam	"Arsenal", "Tottenham", "Luc Alphand Aventures"
other-astronomything	"Algol", "`` Caput Larvae ''", "Zodiac"
other-award	"Grand Commander of the Order of the Niger", "Order of the Republic of Guinea and Nigeria", "GCON"
other-biologything	"BAR", "N-terminal lipid", "Amphiphysin"
other-chemicalthing	"uranium", "carbon dioxide", "sulfur"
other-currency	"lac crore", "$", "Travancore Rupee"
other-disease	"French Dysentery Epidemic of 1779", "hypothyroidism", "bladder cancer"
other-educationaldegree	"Bachelor", "Master", "BSc ( Hons ) in physics"
other-god	"El", "Fujin", "Raijin"
other-language	"English", "Latin", "Breton-speaking"
other-law	"Thirty Years ' Peace", "Leahy–Smith America Invents Act ( AIA", "United States Freedom Support Act"
other-livingthing	"monkeys", "insects", "patchouli"
other-medical	"amitriptyline", "Pediatrics", "pediatrician"
person-actor	"Ellaline Terriss", "Edmund Payne", "Tchéky Karyo"
person-artist/author	"Hicks", "George Axelrod", "Gaetano Donizett"
person-athlete	"Jaguar", "Tozawa", "Neville"
person-director	"Bob Swaim", "Richard Quine", "Frank Darabont"
person-other	"Richard Benson", "Holden", "Campbell"
person-politician	"Emeric", "William", "Rivière"
person-scholar	"Stalmine", "Wurdack", "Stedman"
person-soldier	"Krukenberg", "Joachim Ziegler", "Helmuth Weidling"
product-airplane	"EC135T2 CPDS", "Spey-equipped FGR.2s", "Luton"
product-car	"100EX", "Phantom", "Corvettes - GT1 C6R"
product-food	"red grape", "V. labrusca", "yakiniku"
product-game	"Splinter Cell", "Hardcore RPG", "Airforce Delta"
product-other	"X11", "PDP-1", "Fairbottom Bobs"
product-ship	"HMS `` Chinkara ''", "Essex", "Congress"
product-software	"Wikipedia", "AmiPDF", "Apdf"
product-train	"55022", "Royal Scots Grey", "High Speed Trains"
product-weapon	"ZU-23-2M Wróbel", "ZU-23-2MR Wróbel II", "AR-15 's"

Uses

Direct Use for Inference

python

from span_marker import SpanMarkerModel

# Download from the 🤗 Hub
model = SpanMarkerModel.from_pretrained("guishe/span-marker-bge-base-en-v1.5-fewnerd-fine-super")
# Run inference
entities = model.predict("Most of the Steven Seagal movie \"Under Siege \"(co-starring Tommy Lee Jones) was filmed on the, which is docked on Mobile Bay at Battleship Memorial Park and open to the public.")

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("guishe/span-marker-bge-base-en-v1.5-fewnerd-fine-super")

# 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("guishe/span-marker-bge-base-en-v1.5-fewnerd-fine-super-finetuned")

Training Details

Training Set Metrics

Training set	Min	Median	Max
Sentence length	1	24.4945	267
Entities per sentence	0	2.5832	88

Training Hyperparameters

learning_rate: 1e-05
train_batch_size: 32
eval_batch_size: 32
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.1
num_epochs: 3

Training Results

Epoch	Step	Validation Loss	Validation Precision	Validation Recall	Validation F1	Validation Accuracy
0.5964	3000	0.0324	0.6263	0.5826	0.6037	0.8981
1.1928	6000	0.0278	0.6620	0.6499	0.6559	0.9132
1.7893	9000	0.0264	0.6719	0.6614	0.6666	0.9159
2.3857	12000	0.0260	0.6724	0.6703	0.6714	0.9174
2.9821	15000	0.0258	0.6740	0.6713	0.6726	0.9177

Framework Versions

Python: 3.10.8
SpanMarker: 1.4.0
Transformers: 4.28.0
PyTorch: 1.13.1+cu117
Datasets: 2.14.4
Tokenizers: 0.13.3

Citation

BibTeX

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