metadata
license: mit
base_model: facebook/w2v-bert-2.0
tags:
- generated_from_trainer
metrics:
- wer
model-index:
- name: CiceroASR
results: []
CiceroASR
This model is a fine-tuned version of facebook/w2v-bert-2.0 for the transcription of Classical Latin!
Example from the Aeneid: Transcription: arma virumque cano (Of arms and men I sing)
Example from Genesis: Transcription (little error there): creavit deus chaelum et terram (In the beggining God created the heaven and the earth)
It achieves the following results on the evaluation set of my dataset Latin Youtube:
- Loss: 0.5395
- Wer: 0.2220
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: 0.0001
- train_batch_size: 16
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 300
- num_epochs: 15
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 3.6548 | 0.94 | 50 | 2.8634 | 0.9990 |
| 2.2055 | 1.89 | 100 | 1.0921 | 0.9727 |
| 1.667 | 2.83 | 150 | 0.7201 | 0.4615 |
| 1.3148 | 3.77 | 200 | 0.6431 | 0.3866 |
| 0.9899 | 4.72 | 250 | 0.5561 | 0.3116 |
| 0.9629 | 5.66 | 300 | 0.6027 | 0.3817 |
| 0.7557 | 6.6 | 350 | 0.7145 | 0.3145 |
| 0.9143 | 7.55 | 400 | 0.4926 | 0.2610 |
| 0.5837 | 8.49 | 450 | 0.5396 | 0.2619 |
| 0.7037 | 9.43 | 500 | 0.5076 | 0.2746 |
| 0.5986 | 10.38 | 550 | 0.5224 | 0.2415 |
| 0.5288 | 11.32 | 600 | 0.5332 | 0.2259 |
| 0.5034 | 12.26 | 650 | 0.5436 | 0.2249 |
| 0.4897 | 13.21 | 700 | 0.5171 | 0.2162 |
| 0.4738 | 14.15 | 750 | 0.5395 | 0.2220 |
Framework versions
- Transformers 4.38.1
- Pytorch 2.1.0+cu121
- Datasets 2.17.1
- Tokenizers 0.15.2