metadata
tags:
- generated_from_trainer
datasets:
- evanarlian/common_voice_11_0_id_filtered
metrics:
- wer
model-index:
- name: wav2vec2-xls-r-113m-id
results:
- task:
name: Automatic Speech Recognition
type: automatic-speech-recognition
dataset:
name: evanarlian/common_voice_11_0_id_filtered
type: evanarlian/common_voice_11_0_id_filtered
metrics:
- name: Wer
type: wer
value: 0.6403468314731113
wav2vec2-xls-r-113m-id
This model is a fine-tuned version of evanarlian/distil-wav2vec2-xls-r-113m-id on the evanarlian/common_voice_11_0_id_filtered dataset. It achieves the following results on the evaluation set:
- Loss: 0.5214
- Wer: 0.6403
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: 0.0002
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 16
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.3
- num_epochs: 5.0
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Wer |
|---|---|---|---|---|
| 2.8452 | 0.61 | 1000 | 2.8065 | 1.0 |
| 1.3277 | 1.22 | 2000 | 1.0774 | 0.9330 |
| 1.025 | 1.84 | 3000 | 0.8000 | 0.8474 |
| 0.8497 | 2.45 | 4000 | 0.6812 | 0.7669 |
| 0.7678 | 3.06 | 5000 | 0.6125 | 0.7186 |
| 0.6886 | 3.67 | 6000 | 0.5758 | 0.6812 |
| 0.6318 | 4.29 | 7000 | 0.5420 | 0.6570 |
| 0.6086 | 4.9 | 8000 | 0.5214 | 0.6403 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.12.1
- Datasets 2.7.1
- Tokenizers 0.13.1