Icelandic-finetuned-data-augmentation_light
This model is a fine-tuned version of carlosdanielhernandezmena/wav2vec2-large-xlsr-53-icelandic-ep10-1000h on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2736
- Wer: 0.2338
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: 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: 500
- num_epochs: 10
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
No log | 0.2 | 10 | 0.4566 | 0.2819 |
No log | 0.4 | 20 | 0.3958 | 0.2494 |
No log | 0.6 | 30 | 0.3829 | 0.2237 |
No log | 0.8 | 40 | 0.3622 | 0.2103 |
0.7129 | 1.0 | 50 | 0.3751 | 0.2025 |
0.7129 | 1.2 | 60 | 0.3737 | 0.2025 |
0.7129 | 1.4 | 70 | 0.3765 | 0.2025 |
0.7129 | 1.6 | 80 | 0.3589 | 0.2069 |
0.7129 | 1.8 | 90 | 0.3246 | 0.1902 |
0.3852 | 2.0 | 100 | 0.3146 | 0.1879 |
0.3852 | 2.2 | 110 | 0.3209 | 0.1790 |
0.3852 | 2.4 | 120 | 0.3129 | 0.1779 |
0.3852 | 2.6 | 130 | 0.3003 | 0.1790 |
0.3852 | 2.8 | 140 | 0.2998 | 0.1790 |
0.2803 | 3.0 | 150 | 0.2851 | 0.1868 |
0.2803 | 3.2 | 160 | 0.2753 | 0.1801 |
0.2803 | 3.4 | 170 | 0.2957 | 0.1834 |
0.2803 | 3.6 | 180 | 0.2869 | 0.1790 |
0.2803 | 3.8 | 190 | 0.2650 | 0.1823 |
0.2545 | 4.0 | 200 | 0.2577 | 0.1734 |
0.2545 | 4.2 | 210 | 0.2389 | 0.1779 |
0.2545 | 4.4 | 220 | 0.2330 | 0.1801 |
0.2545 | 4.6 | 230 | 0.2592 | 0.1745 |
0.2545 | 4.8 | 240 | 0.2631 | 0.1779 |
0.2273 | 5.0 | 250 | 0.2305 | 0.1801 |
0.2273 | 5.2 | 260 | 0.2009 | 0.1913 |
0.2273 | 5.4 | 270 | 0.1982 | 0.1946 |
0.2273 | 5.6 | 280 | 0.1849 | 0.2002 |
0.2273 | 5.8 | 290 | 0.2038 | 0.1879 |
0.2192 | 6.0 | 300 | 0.2504 | 0.1857 |
0.2192 | 6.2 | 310 | 0.2993 | 0.1790 |
0.2192 | 6.4 | 320 | 0.2544 | 0.1812 |
0.2192 | 6.6 | 330 | 0.2471 | 0.1969 |
0.2192 | 6.8 | 340 | 0.2688 | 0.1868 |
0.232 | 7.0 | 350 | 0.2264 | 0.2069 |
0.232 | 7.2 | 360 | 0.2695 | 0.1924 |
0.232 | 7.4 | 370 | 0.2728 | 0.1946 |
0.232 | 7.6 | 380 | 0.2508 | 0.1902 |
0.232 | 7.8 | 390 | 0.2499 | 0.1723 |
0.216 | 8.0 | 400 | 0.2035 | 0.1946 |
0.216 | 8.2 | 410 | 0.2620 | 0.1767 |
0.216 | 8.4 | 420 | 0.2655 | 0.1879 |
0.216 | 8.6 | 430 | 0.2773 | 0.2069 |
0.216 | 8.8 | 440 | 0.3075 | 0.2058 |
0.207 | 9.0 | 450 | 0.2791 | 0.1980 |
0.207 | 9.2 | 460 | 0.2045 | 0.1924 |
0.207 | 9.4 | 470 | 0.2329 | 0.2036 |
0.207 | 9.6 | 480 | 0.2200 | 0.2114 |
0.207 | 9.8 | 490 | 0.2864 | 0.2237 |
0.2199 | 10.0 | 500 | 0.2736 | 0.2338 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.0+cu121
- Datasets 2.15.0
- Tokenizers 0.15.0
- Downloads last month
- 4
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social
visibility and check back later, or deploy to Inference Endpoints (dedicated)
instead.