whisper-small-sw-common-voice-ndizi-248
This model is a fine-tuned version of smutuvi/whisper-small-sw-common-voice on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.3100
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: 1e-05
- train_batch_size: 8
- eval_batch_size: 8
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 50
- num_epochs: 100
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.6298 | 1.0 | 28 | 1.2171 |
| 1.1514 | 2.0 | 56 | 1.0364 |
| 0.9175 | 3.0 | 84 | 0.9578 |
| 0.6885 | 4.0 | 112 | 0.9664 |
| 0.5841 | 5.0 | 140 | 1.0001 |
| 0.3397 | 6.0 | 168 | 1.0233 |
| 0.3166 | 7.0 | 196 | 1.0291 |
| 0.2313 | 8.0 | 224 | 1.0749 |
| 0.1457 | 9.0 | 252 | 1.0857 |
| 0.1036 | 10.0 | 280 | 1.0689 |
| 0.0644 | 11.0 | 308 | 1.1099 |
| 0.072 | 12.0 | 336 | 1.1080 |
| 0.0519 | 13.0 | 364 | 1.1119 |
| 0.0312 | 14.0 | 392 | 1.1747 |
| 0.0331 | 15.0 | 420 | 1.1441 |
| 0.02 | 16.0 | 448 | 1.1413 |
| 0.017 | 17.0 | 476 | 1.1880 |
| 0.0157 | 18.0 | 504 | 1.1564 |
| 0.0146 | 19.0 | 532 | 1.1627 |
| 0.013 | 20.0 | 560 | 1.2088 |
| 0.0071 | 21.0 | 588 | 1.2054 |
| 0.006 | 22.0 | 616 | 1.2113 |
| 0.0066 | 23.0 | 644 | 1.2269 |
| 0.0073 | 24.0 | 672 | 1.1721 |
| 0.0064 | 25.0 | 700 | 1.1878 |
| 0.0084 | 26.0 | 728 | 1.1701 |
| 0.0024 | 27.0 | 756 | 1.2221 |
| 0.0056 | 28.0 | 784 | 1.2072 |
| 0.005 | 29.0 | 812 | 1.1742 |
| 0.0032 | 30.0 | 840 | 1.1930 |
| 0.0021 | 31.0 | 868 | 1.1996 |
| 0.0008 | 32.0 | 896 | 1.2344 |
| 0.0014 | 33.0 | 924 | 1.2153 |
| 0.0018 | 34.0 | 952 | 1.2324 |
| 0.0013 | 35.0 | 980 | 1.2281 |
| 0.0011 | 36.0 | 1008 | 1.2223 |
| 0.0006 | 37.0 | 1036 | 1.2326 |
| 0.0011 | 38.0 | 1064 | 1.2250 |
| 0.0007 | 39.0 | 1092 | 1.2270 |
| 0.001 | 40.0 | 1120 | 1.2226 |
| 0.0017 | 41.0 | 1148 | 1.2255 |
| 0.0011 | 42.0 | 1176 | 1.2175 |
| 0.0011 | 43.0 | 1204 | 1.2302 |
| 0.0025 | 44.0 | 1232 | 1.2176 |
| 0.0021 | 45.0 | 1260 | 1.2450 |
| 0.0016 | 46.0 | 1288 | 1.3209 |
| 0.0023 | 47.0 | 1316 | 1.2245 |
| 0.0021 | 48.0 | 1344 | 1.2601 |
| 0.0024 | 49.0 | 1372 | 1.2703 |
| 0.002 | 50.0 | 1400 | 1.2674 |
| 0.0011 | 51.0 | 1428 | 1.2644 |
| 0.0032 | 52.0 | 1456 | 1.2901 |
| 0.0007 | 53.0 | 1484 | 1.2652 |
| 0.0033 | 54.0 | 1512 | 1.2901 |
| 0.0009 | 55.0 | 1540 | 1.2584 |
| 0.0012 | 56.0 | 1568 | 1.2542 |
| 0.0013 | 57.0 | 1596 | 1.2607 |
| 0.0006 | 58.0 | 1624 | 1.2733 |
| 0.0004 | 59.0 | 1652 | 1.2763 |
| 0.0003 | 60.0 | 1680 | 1.2780 |
| 0.0003 | 61.0 | 1708 | 1.2799 |
| 0.0003 | 62.0 | 1736 | 1.2808 |
| 0.0003 | 63.0 | 1764 | 1.2821 |
| 0.0003 | 64.0 | 1792 | 1.2844 |
| 0.0003 | 65.0 | 1820 | 1.2863 |
| 0.0003 | 66.0 | 1848 | 1.2875 |
| 0.0003 | 67.0 | 1876 | 1.2888 |
| 0.0003 | 68.0 | 1904 | 1.2910 |
| 0.0002 | 69.0 | 1932 | 1.2919 |
| 0.0002 | 70.0 | 1960 | 1.2930 |
| 0.0002 | 71.0 | 1988 | 1.2947 |
| 0.0002 | 72.0 | 2016 | 1.2955 |
| 0.0002 | 73.0 | 2044 | 1.2967 |
| 0.0002 | 74.0 | 2072 | 1.2974 |
| 0.0002 | 75.0 | 2100 | 1.2989 |
| 0.0002 | 76.0 | 2128 | 1.2997 |
| 0.0002 | 77.0 | 2156 | 1.3006 |
| 0.0002 | 78.0 | 2184 | 1.3011 |
| 0.0002 | 79.0 | 2212 | 1.3019 |
| 0.0002 | 80.0 | 2240 | 1.3029 |
| 0.0002 | 81.0 | 2268 | 1.3035 |
| 0.0002 | 82.0 | 2296 | 1.3040 |
| 0.0002 | 83.0 | 2324 | 1.3050 |
| 0.0002 | 84.0 | 2352 | 1.3056 |
| 0.0002 | 85.0 | 2380 | 1.3057 |
| 0.0002 | 86.0 | 2408 | 1.3065 |
| 0.0002 | 87.0 | 2436 | 1.3066 |
| 0.0002 | 88.0 | 2464 | 1.3078 |
| 0.0002 | 89.0 | 2492 | 1.3075 |
| 0.0002 | 90.0 | 2520 | 1.3080 |
| 0.0002 | 91.0 | 2548 | 1.3083 |
| 0.0002 | 92.0 | 2576 | 1.3091 |
| 0.0002 | 93.0 | 2604 | 1.3091 |
| 0.0002 | 94.0 | 2632 | 1.3091 |
| 0.0002 | 95.0 | 2660 | 1.3097 |
| 0.0002 | 96.0 | 2688 | 1.3098 |
| 0.0002 | 97.0 | 2716 | 1.3102 |
| 0.0002 | 98.0 | 2744 | 1.3102 |
| 0.0002 | 99.0 | 2772 | 1.3099 |
| 0.0002 | 100.0 | 2800 | 1.3100 |
Framework versions
- Transformers 4.37.0.dev0
- Pytorch 2.1.1+cu121
- Datasets 2.16.1
- Tokenizers 0.15.0