File size: 3,258 Bytes
6f57a7c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
c45bbb0
6f57a7c
 
 
 
 
 
 
 
 
 
 
 
c45bbb0
 
 
945fe26
 
 
 
 
 
 
6f57a7c
 
 
 
 
 
 
3c32998
 
6f57a7c
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
---
license: apache-2.0
base_model: bert-base-cased
tags:
- generated_from_trainer
datasets:
- conll2002
metrics:
- precision
- recall
- f1
- accuracy
model-index:
- name: bert-base-cased-finetuned-conll2002
  results:
  - task:
      name: Token Classification
      type: token-classification
    dataset:
      name: conll2002
      type: conll2002
      config: es
      split: validation
      args: es
    metrics:
    - name: Precision
      type: precision
      value: 0.8175829168559745
    - name: Recall
      type: recall
      value: 0.8269761029411765
    - name: F1
      type: f1
      value: 0.8222526844870915
    - name: Accuracy
      type: accuracy
      value: 0.9739999622092474
---

<!-- 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. -->

# bert-base-cased-finetuned-conll2002

This model is a fine-tuned version of [bert-base-cased](https://huggingface.co/bert-base-cased) on the conll2002 dataset.
It achieves the following results on the evaluation set:
- Loss: 0.1670
- Precision: 0.8176
- Recall: 0.8270
- F1: 0.8223
- Accuracy: 0.9740

## Model description

The model described here is a fine-tuned version of the BERT (Bidirectional Encoder Representations from Transformers) base cased model for 
Named Entity Recognition (NER) tasks, trained on the CoNLL-2002 dataset. BERT is a pre-trained language model based on the transformer architecture, 
designed to understand and process text by considering the context of each word from both directions (left-to-right and right-to-left).

By fine-tuning the BERT base cased model on the CoNLL-2002 dataset, this model has been adapted to recognize and classify named entities such as persons, 
organizations, locations, and other miscellaneous entities within Spanish text. The fine-tuning process involves adjusting the pre-trained model 
weights to better fit the specific task of NER, thereby improving its performance and accuracy on Spanish text.

## Intended uses & limitations

More information needed

## Training and evaluation data

The training was performed using a GPU with 22.5 GB of RAM, 53 GB of system RAM, and 200 GB of disk space. 
This setup ensured efficient handling of the large dataset and the computational demands of fine-tuning the model.

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 2e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 3

### Training results

| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1     | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
| 0.0248        | 1.0   | 1041 | 0.1439          | 0.8066    | 0.8155 | 0.8110 | 0.9732   |
| 0.0141        | 2.0   | 2082 | 0.1569          | 0.8108    | 0.8182 | 0.8145 | 0.9728   |
| 0.0109        | 3.0   | 3123 | 0.1670          | 0.8176    | 0.8270 | 0.8223 | 0.9740   |


### Framework versions

- Transformers 4.41.2
- Pytorch 2.3.0+cu121
- Datasets 2.19.2
- Tokenizers 0.19.1