metadata
license: apache-2.0
base_model: numind/NuNER-v1.0
tags:
- token-classification
- ner
- named-entity-recognition
metrics:
- precision
- recall
- f1
- accuracy
model-index:
- name: nuner-v1_ontonotes5
results:
- task:
type: token-classification
name: Named Entity Recognition
dataset:
name: OntoNotes5
type: tner/ontonotes5
split: eval
metrics:
- type: f1
value: 0.890930568316052
name: F1
- type: precision
value: 0.8777586206896552
name: Precision
- type: recall
value: 0.9045038642622368
name: Recall
- type: accuracy
value: 0.9818887790313181
name: Accuracy
datasets:
- tner/ontonotes5
language:
- en
library_name: transformers
pipeline_tag: token-classification
widget:
- text: >-
Concern and scepticism surround Niger uranium mining waste storage plans.
Towering mounds dot the desert landscape in northern Niger's Arlit region,
but they are heaps of partially radioactive waste left from four decades
of operations at one of the world's biggest uranium mines. An ambitious
10-year scheme costing $160 million is underway to secure the waste and
avoid risks to health and the environment, but many local people are
worried or sceptical. France's nuclear giant Areva, now called Orano,
worked the area under a subsidiary, the Akouta Mining Company (Cominak).
Cominak closed the site in 2021 after extracting 75,000 tonnes of uranium,
much of which went to fuelling the scores of nuclear reactors that provide
the backbone of France's electricity supply. Cominak's director general
Mahaman Sani Abdoulaye showcased the rehabilitation project to the first
French journalists to visit the site since 2010, when seven Areva
employees were kidnapped by jihadists.
- text: >-
SE Michigan counties allege insulin gouging; Localities file lawsuit
against pharmaceutical makers. Four metro Detroit counties filed federal
lawsuits Wednesday against some of the nation's biggest pharmaceutical
manufacturers and pharmacy benefit managers alleging illegal price fixing
for insulin products. Macomb, Monroe, Wayne and Washtenaw counties filed
the lawsuits in U.S. District Court in New Jersey against more than a
dozen companies, including Lilly, Sanofi Aventis, Novo Nordisk, Express
Scripts, Optum Rx and CVS Caremark, per their attorneys. "These are the
first such lawsuits that have been filed in the state of Michigan and
probably more to come," said attorney Melvin Butch Hollowell of the Miller
Law Firm. He described the allegations during a news conference, saying
that nationally "the pharmacies and manufacturers get together. They
control about 90% of the market each, of the insulin market. They talk to
each other secretly. And they jack up the prices through anticompetitive
means. And what we've seen is over the past 20 years, when we talk about
jacking up the prices, they jack them up 1,500% in the last 20 years.
1,500%."
- text: >-
Foreign governments may be spying on your smartphone notifications,
senator says. Washington (CNN) — Foreign governments have reportedly
attempted to spy on iPhone and Android users through the mobile app
notifications they receive on their smartphones - and the US government
has forced Apple and Google to keep quiet about it, according to a top US
senator. Through legal demands sent to the tech giants, governments have
allegedly tried to force Apple and Google to turn over sensitive
information that could include the contents of a notification - such as
previews of a text message displayed on a lock screen, or an update about
app activity, Oregon Democratic Sen. Ron Wyden said in a new report.
Wyden's report reflects the latest example of long-running tensions
between tech companies and governments over law enforcement demands, which
have stretched on for more than a decade. Governments around the world
have particularly battled with tech companies over encryption, which
provides critical protections to users and businesses while in some cases
preventing law enforcement from pursuing investigations into messages sent
over the internet.
- text: >-
Tech giants ‘could severely disable UK spooks from stopping online harms’.
Silicon Valley tech giants’ actions could “severely disable” UK spooks
from preventing harm caused by online paedophiles and fraudsters, Suella
Braverman has suggested. The Conservative former home secretary named
Facebook owner Meta , and Apple, and their use of technologies such as
end-to-end encryption as a threat to attempts to tackle digital crimes.
She claimed the choice to back these technologies without “safeguards”
could “enable and indeed facilitate some of the worst atrocities that our
brave men and women in law enforcement agencies deal with every day”, as
MPs began considering changes to investigatory powers laws. The
Investigatory Powers (Amendment) Bill includes measures to make it easier
for agencies to examine and retain bulk datasets, such as publicly
available online telephone records, and would allow intelligence agencies
to use internet connection records to aid detection of their targets. We
know that the terrorists, the serious organised criminals, and fraudsters,
and the online paedophiles, all take advantage of the dark web and
encrypted spaces
- text: >-
Camargo Corrêa asks Toffoli to suspend the fine agreed with Lava Jato. The
Camargo Corrêa group has asked Justice Dias Toffoli to suspend the R$1.4
billion fine it agreed to pay in its leniency agreement under Operation
Car Wash. The company asked for an extension of the minister's decisions
that benefited J&F and Odebrecht. Like the other companies, it claimed
that it suffered undue pressure from members of the Federal Public
Prosecutor's Office (MPF) to close the deal. Much of the request is based
on messages exchanged between prosecutors from the Curitiba task force and
former judge Sergio Moro - Camargo Corrêa requested full access to the
material, seized in Operation Spoofing, which arrested the hackers who
broke into cell phones. The dialogues, according to the group's defense,
indicate that the executives did not freely agree to the deal, since they
were the targets of lawsuits and pre-trial detentions.
numind/NuNER-v1.0 fine-tuned on OntoNotes5
This is a NuNER model fine-tuned on the OntoNotes5 dataset that can be used for Named Entity Recognition. NuNER model uses RoBERTa-base as the backbone encoder and it was trained on the NuNER dataset, which is a large and diverse dataset synthetically labeled by gpt-3.5-turbo-0301 of 1M sentences. This further pre-training phase allowed the generation of high quality token embeddings, a good starting point for fine-tuning on more specialized datasets.
Model Details
The model was fine-tuned as a regular BERT-based model for NER task using HuggingFace Trainer class.
Model labels
Entity Types: CARDINAL, DATE, PERSON, NORP, GPE, LAW, PERCENT, ORDINAL, MONEY, WORK_OF_ART, FAC, TIME, QUANTITY, PRODUCT, LANGUAGE, ORG, LOC, EVENT
Uses
Direct Use for Inference
>>> from transformers import pipeline
>>> text = """Foreign governments may be spying on your smartphone notifications, senator says. Washington (CNN) — Foreign governments have reportedly attempted to spy on iPhone and Android users through the mobile app notifications they receive on their smartphones - and the US government has forced Apple and Google to keep quiet about it, according to a top US senator. Through legal demands sent to the tech giants, governments have allegedly tried to force Apple and Google to turn over sensitive information that could include the contents of a notification - such as previews of a text message displayed on a lock screen, or an update about app activity, Oregon Democratic Sen. Ron Wyden said in a new report. Wyden's report reflects the latest example of long-running tensions between tech companies and governments over law enforcement demands, which have stretched on for more than a decade. Governments around the world have particularly battled with tech companies over encryption, which provides critical protections to users and businesses while in some cases preventing law enforcement from pursuing investigations into messages sent over the internet."""
>>> classifier = pipeline(
"ner",
model="guishe/nuner-v1_ontonotes5",
aggregation_strategy="simple",
)
>>> classifier(text)
[{'entity_group': 'GPE',
'score': 0.99179757,
'word': ' Washington',
'start': 82,
'end': 92},
{'entity_group': 'ORG',
'score': 0.9535868,
'word': 'CNN',
'start': 94,
'end': 97},
{'entity_group': 'PRODUCT',
'score': 0.6833637,
'word': ' iPhone',
'start': 157,
'end': 163},
{'entity_group': 'PRODUCT',
'score': 0.5540275,
'word': ' Android',
'start': 168,
'end': 175},
{'entity_group': 'GPE',
'score': 0.98848885,
'word': ' US',
'start': 263,
'end': 265},
{'entity_group': 'ORG',
'score': 0.9939406,
'word': ' Apple',
'start': 288,
'end': 293},
{'entity_group': 'ORG',
'score': 0.9933014,
'word': ' Google',
'start': 298,
'end': 304},
{'entity_group': 'GPE',
'score': 0.99083686,
'word': ' US',
'start': 348,
'end': 350},
{'entity_group': 'ORG',
'score': 0.99349517,
'word': ' Apple',
'start': 449,
'end': 454},
{'entity_group': 'ORG',
'score': 0.99239254,
'word': ' Google',
'start': 459,
'end': 465},
{'entity_group': 'GPE',
'score': 0.99598336,
'word': ' Oregon',
'start': 649,
'end': 655},
{'entity_group': 'NORP',
'score': 0.99030787,
'word': ' Democratic',
'start': 656,
'end': 666},
{'entity_group': 'PERSON',
'score': 0.9957912,
'word': ' Ron Wyden',
'start': 672,
'end': 681},
{'entity_group': 'PERSON',
'score': 0.83941424,
'word': ' Wyden',
'start': 704,
'end': 709},
{'entity_group': 'DATE',
'score': 0.87746465,
'word': ' more than a decade',
'start': 869,
'end': 887}]
Training Details
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 64
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 4
Training results
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
|---|---|---|---|---|---|---|---|
| 0.0781 | 1.0 | 936 | 0.0754 | 0.8392 | 0.8843 | 0.8612 | 0.9778 |
| 0.049 | 2.0 | 1873 | 0.0685 | 0.8597 | 0.8935 | 0.8763 | 0.9794 |
| 0.0357 | 3.0 | 2809 | 0.0714 | 0.8608 | 0.9016 | 0.8807 | 0.9806 |
| 0.027 | 4.0 | 3744 | 0.0728 | 0.8712 | 0.9000 | 0.8853 | 0.9811 |
Framework versions
BibTeX
@misc{bogdanov2024nuner,
title={NuNER: Entity Recognition Encoder Pre-training via LLM-Annotated Data},
author={Sergei Bogdanov and Alexandre Constantin and Timothée Bernard and Benoit Crabbé and Etienne Bernard},
year={2024},
eprint={2402.15343},
archivePrefix={arXiv},
primaryClass={cs.CL}
}