|
--- |
|
license: apache-2.0 |
|
tags: |
|
- generated_from_trainer |
|
base_model: bert-base-uncased |
|
datasets: |
|
- conll2003 |
|
metrics: |
|
- precision |
|
- recall |
|
- f1 |
|
- accuracy |
|
model-index: |
|
- name: NER_Pittsburgh_TAA |
|
results: |
|
- task: |
|
type: token-classification |
|
name: Token Classification |
|
dataset: |
|
name: conll2003 |
|
type: conll2003 |
|
config: conll2003 |
|
split: validation |
|
args: conll2003 |
|
metrics: |
|
- type: precision |
|
value: 0.9429236395877203 |
|
name: Precision |
|
- type: recall |
|
value: 0.9517843159190066 |
|
name: Recall |
|
- type: f1 |
|
value: 0.9473332591025497 |
|
name: F1 |
|
- type: accuracy |
|
value: 0.9867030994328562 |
|
name: Accuracy |
|
language: |
|
- en |
|
- uk |
|
--- |
|
|
|
<!-- This model card has been generated automatically according to the information the Trainer had access to. You |
|
should probably proofread and complete it, then remove this comment. --> |
|
|
|
# NER_Pittsburgh_TAA |
|
|
|
This model is a fine-tuned version of [bert-base-uncased](https://huggingface.co/bert-base-uncased) on the conll2003 dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.0860 |
|
- Precision: 0.9429 |
|
- Recall: 0.9518 |
|
- F1: 0.9473 |
|
- Accuracy: 0.9867 |
|
|
|
## Model description |
|
|
|
## Ukr |
|
|
|
Модель була створена як практичне завдання з машиного навчання, це за fine-tuning BERT модель для задачі Named Entity Recognition. |
|
Датасет який був використан це conll2003, стандат для навчання моделей під задачу Named Entity Recognition, або ще визначення складових мови в реченні. |
|
Дізнатися як працює модель маєте змогу або через інтерфейс, який надає huggingface, або ж через код |
|
|
|
from transformers import AutoTokenizer, AutoModelForTokenClassification |
|
|
|
tokenizer = AutoTokenizer.from_pretrained("CineAI/NER_Pittsburgh_TAA") |
|
|
|
model = AutoModelForTokenClassification.from_pretrained("CineAI/NER_Pittsburgh_TAA") |
|
|
|
Якщо цікавить чому модель має таку назву, перше це для чого вона для NER, друга складова це назва крутої пісні Pittsburgh третя і остання складова |
|
це гурт який пісню створив це The Amity Affliction |
|
|
|
## En |
|
|
|
The model was created as a practical machine learning task, it is a fine-tuning BERT model for the Named Entity Recognition task. |
|
The dataset used is conll2003, a standard for training models for the Named Entity Recognition task, or for identifying the components of speech in a sentence. |
|
You can find out how the model works either through the interface provided by huggingface or through the code |
|
|
|
from transformers import AutoTokenizer, AutoModelForTokenClassification |
|
|
|
tokenizer = AutoTokenizer.from_pretrained("CineAI/NER_Pittsburgh_TAA") |
|
|
|
model = AutoModelForTokenClassification.from_pretrained("CineAI/NER_Pittsburgh_TAA") |
|
|
|
If you are wondering why the model has such a name, the first is why it is for NER, the second component is the name of a cool song Pittsburgh, the third and last component |
|
is the band that created the song - The Amity Affliction |
|
|
|
## Intended uses & limitations |
|
|
|
Everyone can use this model, it is completely free and distributed under the Apache 2.0 licence. |
|
|
|
## Training and evaluation data |
|
|
|
Training and assessment data are the same - conll2003 |
|
|
|
## Training procedure |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 2e-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 |
|
- num_epochs: 5 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy | |
|
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:| |
|
| No log | 1.0 | 439 | 0.0863 | 0.9437 | 0.9444 | 0.9440 | 0.9861 | |
|
| 0.0024 | 2.0 | 878 | 0.0995 | 0.9394 | 0.9442 | 0.9418 | 0.9852 | |
|
| 0.0021 | 3.0 | 1317 | 0.0904 | 0.9355 | 0.9463 | 0.9409 | 0.9856 | |
|
| 0.0012 | 4.0 | 1756 | 0.0835 | 0.9427 | 0.9514 | 0.9471 | 0.9867 | |
|
| 0.0009 | 5.0 | 2195 | 0.0860 | 0.9429 | 0.9518 | 0.9473 | 0.9867 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.41.2 |
|
- Pytorch 2.3.0+cu121 |
|
- Datasets 2.19.2 |
|
- Tokenizers 0.19.1 |
