---
library_name: transformers
language:
- lg
license: apache-2.0
base_model: google-bert/bert-base-multilingual-cased
tags:
- named-entity-recognition
- luganda
- african-languages
- pii-detection
- token-classification
- generated_from_trainer
datasets:
- Beijuka/Luganda_Monolingual_PII_dataset
metrics:
- precision
- recall
- f1
- accuracy
model-index:
- name: luganda-ner-bert-v7
  results:
  - task:
      name: Token Classification
      type: token-classification
    dataset:
      name: Beijuka/Luganda_Monolingual_PII_dataset
      type: Beijuka/Luganda_Monolingual_PII_dataset
      args: 'split: train+validation+test'
    metrics:
    - name: Precision
      type: precision
      value: 0.8336713995943205
    - name: Recall
      type: recall
      value: 0.8162859980139027
    - name: F1
      type: f1
      value: 0.8248871048670346
    - name: Accuracy
      type: accuracy
      value: 0.9459025442866462
---

<!-- 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. -->

# luganda-ner-bert-v7

This model is a fine-tuned version of [google-bert/bert-base-multilingual-cased](https://huggingface.co/google-bert/bert-base-multilingual-cased) on the Beijuka/Luganda_Monolingual_PII_dataset dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3195
- Precision: 0.8337
- Recall: 0.8163
- F1: 0.8249
- Accuracy: 0.9459

## Model description

More information needed

## Intended uses & limitations

More information needed

## Training and evaluation data

More information needed

## Training procedure

### Training hyperparameters

The following hyperparameters were used during training:
- learning_rate: 5e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: linear
- num_epochs: 10

### Training results

| Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1     | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:---------:|:------:|:------:|:--------:|
| No log        | 1.0   | 261  | 0.4382          | 0.5831    | 0.5124 | 0.5455 | 0.8761   |
| 0.4959        | 2.0   | 522  | 0.2885          | 0.7147    | 0.7289 | 0.7217 | 0.9171   |
| 0.4959        | 3.0   | 783  | 0.2778          | 0.7592    | 0.7200 | 0.7390 | 0.9308   |
| 0.1547        | 4.0   | 1044 | 0.2552          | 0.8016    | 0.7865 | 0.7940 | 0.9389   |
| 0.1547        | 5.0   | 1305 | 0.2718          | 0.7908    | 0.7885 | 0.7897 | 0.9389   |
| 0.061         | 6.0   | 1566 | 0.2881          | 0.8360    | 0.7795 | 0.8068 | 0.9426   |
| 0.061         | 7.0   | 1827 | 0.2908          | 0.8008    | 0.8143 | 0.8075 | 0.9446   |
| 0.023         | 8.0   | 2088 | 0.3195          | 0.8337    | 0.8163 | 0.8249 | 0.9459   |
| 0.023         | 9.0   | 2349 | 0.3190          | 0.8208    | 0.8143 | 0.8175 | 0.9476   |
| 0.0088        | 10.0  | 2610 | 0.3318          | 0.8354    | 0.7964 | 0.8155 | 0.9458   |


### Framework versions

- Transformers 4.53.0
- Pytorch 2.6.0+cu124
- Datasets 3.6.0
- Tokenizers 0.21.2