|
--- |
|
license: apache-2.0 |
|
base_model: facebook/wav2vec2-xls-r-1b |
|
tags: |
|
- generated_from_trainer |
|
datasets: |
|
- common_voice_13_0 |
|
metrics: |
|
- wer |
|
model-index: |
|
- name: LugandaASRwav20Vec1B |
|
results: |
|
- task: |
|
name: Automatic Speech Recognition |
|
type: automatic-speech-recognition |
|
dataset: |
|
name: common_voice_13_0 |
|
type: common_voice_13_0 |
|
config: lg |
|
split: validation |
|
args: lg |
|
metrics: |
|
- name: Wer |
|
type: wer |
|
value: 0.23043478260869565 |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# LugandaASRwav20Vec1B |
|
|
|
This model is a fine-tuned version of [facebook/wav2vec2-xls-r-1b](https://huggingface.co/facebook/wav2vec2-xls-r-1b) on the common_voice_13_0 dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.1854 |
|
- Wer: 0.2304 |
|
|
|
## 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: 0.0003 |
|
- train_batch_size: 4 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 24 |
|
- total_train_batch_size: 96 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- lr_scheduler_warmup_steps: 200 |
|
- num_epochs: 3 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Wer | |
|
|:-------------:|:-----:|:----:|:---------------:|:------:| |
|
| 4.303 | 0.14 | 100 | 2.1141 | 1.0 | |
|
| 0.7155 | 0.27 | 200 | 0.5656 | 0.6752 | |
|
| 0.4493 | 0.41 | 300 | 0.4402 | 0.5607 | |
|
| 0.3964 | 0.54 | 400 | 0.3918 | 0.5114 | |
|
| 0.3646 | 0.68 | 500 | 0.3601 | 0.4592 | |
|
| 0.3294 | 0.81 | 600 | 0.3381 | 0.4467 | |
|
| 0.3339 | 0.95 | 700 | 0.3340 | 0.4266 | |
|
| 0.2893 | 1.08 | 800 | 0.2913 | 0.3670 | |
|
| 0.2743 | 1.22 | 900 | 0.2854 | 0.3600 | |
|
| 0.262 | 1.36 | 1000 | 0.2666 | 0.3318 | |
|
| 0.2545 | 1.49 | 1100 | 0.2601 | 0.3341 | |
|
| 0.2437 | 1.63 | 1200 | 0.2488 | 0.3152 | |
|
| 0.2235 | 1.76 | 1300 | 0.2416 | 0.3015 | |
|
| 0.2188 | 1.9 | 1400 | 0.2330 | 0.2902 | |
|
| 0.2054 | 2.03 | 1500 | 0.2218 | 0.2750 | |
|
| 0.1743 | 2.17 | 1600 | 0.2153 | 0.2672 | |
|
| 0.1722 | 2.3 | 1700 | 0.2098 | 0.2575 | |
|
| 0.1656 | 2.44 | 1800 | 0.2011 | 0.2538 | |
|
| 0.1608 | 2.58 | 1900 | 0.2000 | 0.2475 | |
|
| 0.1574 | 2.71 | 2000 | 0.1937 | 0.2428 | |
|
| 0.1531 | 2.85 | 2100 | 0.1882 | 0.2347 | |
|
| 0.1451 | 2.98 | 2200 | 0.1854 | 0.2304 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.32.0 |
|
- Pytorch 2.0.1+cu118 |
|
- Datasets 2.13.0 |
|
- Tokenizers 0.13.3 |
|
|
