metadata
license: apache-2.0
tags:
- generated_from_trainer
datasets:
- xtreme_s
metrics:
- wer
base_model: facebook/wav2vec2-base
model-index:
- name: wav2vec2-XLS-R-Fleurs-demo-google-colab-Ezra_William_Prod2
results:
- task:
type: automatic-speech-recognition
name: Automatic Speech Recognition
dataset:
name: xtreme_s
type: xtreme_s
config: fleurs.id_id
split: test
args: fleurs.id_id
metrics:
- type: wer
value: 1
name: Wer
wav2vec2-XLS-R-Fleurs-demo-google-colab-Ezra_William_Prod2
This model is a fine-tuned version of facebook/wav2vec2-base on the xtreme_s dataset. It achieves the following results on the evaluation set:
- Loss: 2.8837
- Wer: 1.0
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: 0.005
- 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_steps: 200
- num_epochs: 60
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
3.962 | 3.08 | 100 | 2.8983 | 1.0 |
2.9085 | 6.15 | 200 | 2.8864 | 1.0 |
2.9094 | 9.23 | 300 | 2.9040 | 1.0 |
2.8976 | 12.31 | 400 | 2.9628 | 1.0 |
2.901 | 15.38 | 500 | 2.8694 | 1.0 |
2.8913 | 18.46 | 600 | 2.8954 | 1.0 |
2.8918 | 21.54 | 700 | 2.8726 | 1.0 |
2.892 | 24.62 | 800 | 2.8865 | 1.0 |
2.8856 | 27.69 | 900 | 2.9127 | 1.0 |
2.8893 | 30.77 | 1000 | 2.8989 | 1.0 |
2.8862 | 33.85 | 1100 | 2.8831 | 1.0 |
2.8853 | 36.92 | 1200 | 2.8960 | 1.0 |
2.8856 | 40.0 | 1300 | 2.8911 | 1.0 |
2.8849 | 43.08 | 1400 | 2.8926 | 1.0 |
2.8829 | 46.15 | 1500 | 2.8837 | 1.0 |
2.8812 | 49.23 | 1600 | 2.8859 | 1.0 |
2.8825 | 52.31 | 1700 | 2.8858 | 1.0 |
2.8833 | 55.38 | 1800 | 2.8856 | 1.0 |
2.88 | 58.46 | 1900 | 2.8837 | 1.0 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.2+cu121
- Datasets 1.18.3
- Tokenizers 0.15.1