metadata
license: apache-2.0
base_model: facebook/wav2vec2-base
tags:
- generated_from_trainer
metrics:
- accuracy
model-index:
- name: wave2vec2_capstone
results: []
datasets:
- mozilla-foundation/common_voice_16_1
language:
- en
- ca
- rw
- be
- eo
- de
- fr
- ka
- es
- lg
- sw
- fa
- it
- mh
- zh
- ba
- ta
- ru
- eu
- th
- pt
- pl
- ja
pipeline_tag: audio-classification
wave2vec2_capstone
This model is a fine-tuned version of facebook/wav2vec2-base on the common_voice_16_1 dataset. It achieves the following results on the evaluation set:
- Loss: 0.2796
- Accuracy: 0.9400
- F1 score: 0.9399
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: 9
- eval_batch_size: 9
- seed: 42
- gradient_accumulation_steps: 12
- total_train_batch_size: 108
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 8
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 score |
|---|---|---|---|---|---|
| 0.8951 | 1.0 | 776 | 1.1617 | 0.6651 | 0.6607 |
| 0.6608 | 2.0 | 1552 | 0.6345 | 0.8188 | 0.8188 |
| 0.4426 | 3.0 | 2328 | 0.4792 | 0.8672 | 0.8677 |
| 0.3576 | 4.0 | 3105 | 0.3826 | 0.8917 | 0.8929 |
| 0.194 | 5.0 | 3881 | 0.3255 | 0.9125 | 0.9130 |
| 0.1635 | 6.0 | 4657 | 0.2903 | 0.9208 | 0.9206 |
| 0.0903 | 7.0 | 5433 | 0.2990 | 0.9300 | 0.9299 |
| 0.0405 | 8.0 | 6208 | 0.2796 | 0.9400 | 0.9399 |
Framework versions
- Transformers 4.36.2
- Pytorch 2.1.2+cu121
- Datasets 2.16.0
- Tokenizers 0.15.0