|
--- |
|
license: apache-2.0 |
|
base_model: facebook/wav2vec2-conformer-rel-pos-large |
|
tags: |
|
- generated_from_trainer |
|
metrics: |
|
- wer |
|
model-index: |
|
- name: wav2vec2-conformer-rel-pos-jv-openslr |
|
results: [] |
|
datasets: |
|
- openslr/openslr |
|
language: |
|
- jv |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# wav2vec2-conformer-rel-pos-jv-openslr |
|
|
|
This model is a fine-tuned version of [facebook/wav2vec2-conformer-rel-pos-large](https://huggingface.co/facebook/wav2vec2-conformer-rel-pos-large) on the None dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.2470 |
|
- Wer: 0.1227 |
|
|
|
## Model description |
|
|
|
The model is a fine-tuned version of wav2vec2-conformer-rel-pos-large, specifically adapted using the OpenSLR 41 dataset, which is focused on the Javanese language domain. This adaptation enables the model to effectively recognize and process spoken Javanese, leveraging the robust capabilities of the wav2vec2-conformer-rel-pos-large architecture combined with domain-specific training data. |
|
|
|
## Intended uses & limitations |
|
|
|
This model is intended for transcribing spoken Javanese language from audio recordings. It achieves a Word Error Rate (WER) of 12%, indicating that while the model performs reasonably well, it still produces significant transcription errors. Users should be aware that the accuracy may vary, particularly in cases with challenging audio conditions or less common dialects. Additionally, this model requires input audio at a sample rate of 16kHz, which may limit its applicability for recordings at different sample rates or lower quality audio files. |
|
|
|
## Training and evaluation data |
|
|
|
The model use OpenSLR41 datasets, and split into 2 section (training and testing), then the model is trained using 1xA100 GPU with a training duration of NaN hours. |
|
|
|
### Training hyperparameters |
|
|
|
The following hyperparameters were used during training: |
|
- learning_rate: 0.0001 |
|
- 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 |
|
- lr_scheduler_warmup_steps: 1000 |
|
- num_epochs: 75 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Wer | |
|
|:-------------:|:-------:|:-----:|:---------------:|:------:| |
|
| 0.5826 | 2.8329 | 2000 | 0.4733 | 0.4445 | |
|
| 0.3478 | 5.6657 | 4000 | 0.3538 | 0.3191 | |
|
| 0.2532 | 8.4986 | 6000 | 0.3085 | 0.2646 | |
|
| 0.2028 | 11.3314 | 8000 | 0.2799 | 0.2467 | |
|
| 0.1628 | 14.1643 | 10000 | 0.2623 | 0.2095 | |
|
| 0.1407 | 16.9972 | 12000 | 0.2510 | 0.2068 | |
|
| 0.1154 | 19.8300 | 14000 | 0.2922 | 0.1937 | |
|
| 0.1044 | 22.6629 | 16000 | 0.2660 | 0.1730 | |
|
| 0.0929 | 25.4958 | 18000 | 0.2818 | 0.1868 | |
|
| 0.0798 | 28.3286 | 20000 | 0.2573 | 0.1633 | |
|
| 0.074 | 31.1615 | 22000 | 0.2398 | 0.1647 | |
|
| 0.0678 | 33.9943 | 24000 | 0.2601 | 0.1606 | |
|
| 0.0628 | 36.8272 | 26000 | 0.2627 | 0.1613 | |
|
| 0.057 | 39.6601 | 28000 | 0.2393 | 0.1468 | |
|
| 0.0547 | 42.4929 | 30000 | 0.2662 | 0.1585 | |
|
| 0.0512 | 45.3258 | 32000 | 0.2544 | 0.1502 | |
|
| 0.0446 | 48.1586 | 34000 | 0.2542 | 0.1502 | |
|
| 0.045 | 50.9915 | 36000 | 0.2624 | 0.1516 | |
|
| 0.0403 | 53.8244 | 38000 | 0.2487 | 0.1420 | |
|
| 0.0378 | 56.6572 | 40000 | 0.2498 | 0.1330 | |
|
| 0.0353 | 59.4901 | 42000 | 0.2495 | 0.1309 | |
|
| 0.0337 | 62.3229 | 44000 | 0.2505 | 0.1316 | |
|
| 0.029 | 65.1558 | 46000 | 0.2373 | 0.1247 | |
|
| 0.0277 | 67.9887 | 48000 | 0.2543 | 0.1282 | |
|
| 0.0283 | 70.8215 | 50000 | 0.2547 | 0.1234 | |
|
| 0.0275 | 73.6544 | 52000 | 0.2470 | 0.1227 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.44.0 |
|
- Pytorch 2.2.1+cu118 |
|
- Datasets 2.20.0 |
|
- Tokenizers 0.19.1 |
