sartajbhuvaji/bert-named-entity-recognition

Model Card for Bert Named Entity Recognition

Model Description

This is a chat fine-tuned version of google-bert/bert-base-uncased, designed to perform Named Entity Recognition on a text sentence imput.

• Developed by: Sartaj
• Finetuned from model: google-bert/bert-base-uncased
• Language(s): English
• License: apache-2.0
• Framework: Hugging Face Transformers

Model Sources

• Repository: google-bert/bert-base-uncased
• Paper: BERT-paper

Uses

Model can be used to recognize Named Entities in text.

Usage

from transformers import AutoTokenizer, AutoModelForTokenClassification
from transformers import pipeline

tokenizer = AutoTokenizer.from_pretrained("sartajbhuvaji/bert-named-entity-recognition")
model = AutoModelForTokenClassification.from_pretrained("sartajbhuvaji/bert-named-entity-recognition")

nlp = pipeline("ner", model=model, tokenizer=tokenizer)
example = "My name is Wolfgang and I live in Berlin"

ner_results = nlp(example)
print(ner_results)

[
  {
    "end": 19,
    "entity": "B-PER",
    "index": 4,
    "score": 0.99633455,
    "start": 11,
    "word": "wolfgang"
  },
  {
    "end": 40,
    "entity": "B-LOC",
    "index": 9,
    "score": 0.9987465,
    "start": 34,
    "word": "berlin"
  }
]

Training Details

• Dataset : eriktks/conll2003

Abbreviation	Description
O	Outside of a named entity
B-MISC	Beginning of a miscellaneous entity right after another miscellaneous entity
I-MISC	Miscellaneous entity
B-PER	Beginning of a person's name right after another person's name
I-PER	Person's name
B-ORG	Beginning of an organization right after another organization
I-ORG	Organization
B-LOC	Beginning of a location right after another location
I-LOC	Location

Training Procedure

• Full Model Finetune
• Epochs : 5

Training Loss Curves

Trainer

• global_step: 4390
• training_loss: 0.040937909830132485
• train_runtime: 206.3611
• train_samples_per_second: 340.205
• train_steps_per_second: 21.273
• total_flos: 1702317283240608.0
• train_loss: 0.040937909830132485
• epoch: 5.0

Evaluation

• Precision: 0.8992
• Recall: 0.9115
• F1 Score: 0.9053

Classification Report

Class	Precision	Recall	F1-Score	Support
LOC	0.91	0.93	0.92	1668
MISC	0.76	0.81	0.78	702
ORG	0.87	0.88	0.88	1661
PER	0.98	0.97	0.97	1617
Micro Avg	0.90	0.91	0.91	5648
Macro Avg	0.88	0.90	0.89	5648
Weighted Avg	0.90	0.91	0.91	5648

• Evaluation Dataset : eriktks/conll2003