<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
wav2vec2-xls-r-300m-arabic_speech_commands_10s
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.2029
- Accuracy: 0.9625
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: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 4
- 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: 0.1
- num_epochs: 80
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy |
---|---|---|---|---|
3.6904 | 1.0 | 25 | 3.6889 | 0.0292 |
3.6892 | 2.0 | 50 | 3.6876 | 0.025 |
3.6902 | 3.0 | 75 | 3.6707 | 0.0792 |
3.6133 | 4.0 | 100 | 3.5175 | 0.0917 |
3.4989 | 5.0 | 125 | 3.3043 | 0.0917 |
3.1971 | 6.0 | 150 | 3.1385 | 0.125 |
3.1348 | 7.0 | 175 | 2.8781 | 0.2 |
2.8271 | 8.0 | 200 | 2.6197 | 0.2417 |
2.7187 | 9.0 | 225 | 2.4002 | 0.2417 |
2.3602 | 10.0 | 250 | 2.1351 | 0.4167 |
2.1853 | 11.0 | 275 | 1.9951 | 0.4042 |
2.0971 | 12.0 | 300 | 2.0668 | 0.3667 |
1.8549 | 13.0 | 325 | 1.7583 | 0.4792 |
1.6929 | 14.0 | 350 | 1.5585 | 0.5542 |
1.4449 | 15.0 | 375 | 1.5602 | 0.5292 |
1.5174 | 16.0 | 400 | 1.4584 | 0.6 |
1.4283 | 17.0 | 425 | 1.3407 | 0.65 |
1.2341 | 18.0 | 450 | 1.1690 | 0.6708 |
1.2353 | 19.0 | 475 | 1.0808 | 0.7542 |
1.0083 | 20.0 | 500 | 0.8947 | 0.7917 |
0.8907 | 21.0 | 525 | 0.9383 | 0.7958 |
0.9161 | 22.0 | 550 | 0.7948 | 0.7917 |
0.781 | 23.0 | 575 | 0.7135 | 0.8375 |
0.7709 | 24.0 | 600 | 0.8259 | 0.8208 |
0.5534 | 25.0 | 625 | 0.6845 | 0.7958 |
0.6089 | 26.0 | 650 | 0.8161 | 0.7875 |
0.5412 | 27.0 | 675 | 0.5962 | 0.8292 |
0.4843 | 28.0 | 700 | 0.5888 | 0.8792 |
0.5755 | 29.0 | 725 | 0.5389 | 0.8417 |
0.4687 | 30.0 | 750 | 0.5176 | 0.8917 |
0.4191 | 31.0 | 775 | 0.4904 | 0.8542 |
0.4361 | 32.0 | 800 | 0.5360 | 0.8875 |
0.264 | 33.0 | 825 | 0.4501 | 0.8958 |
0.3062 | 34.0 | 850 | 0.5384 | 0.8917 |
0.2992 | 35.0 | 875 | 0.4840 | 0.9167 |
0.3904 | 36.0 | 900 | 0.4934 | 0.8792 |
0.2689 | 37.0 | 925 | 0.3123 | 0.925 |
0.1963 | 38.0 | 950 | 0.4691 | 0.8875 |
0.2402 | 39.0 | 975 | 0.4508 | 0.8792 |
0.1912 | 40.0 | 1000 | 0.3873 | 0.8917 |
0.1512 | 41.0 | 1025 | 0.3153 | 0.9208 |
0.1673 | 42.0 | 1050 | 0.2599 | 0.9417 |
0.1981 | 43.0 | 1075 | 0.5351 | 0.8917 |
0.1977 | 44.0 | 1100 | 0.4595 | 0.9042 |
0.2621 | 45.0 | 1125 | 0.2737 | 0.9375 |
0.1239 | 46.0 | 1150 | 0.1870 | 0.95 |
0.2602 | 47.0 | 1175 | 0.2738 | 0.9167 |
0.0681 | 48.0 | 1200 | 0.2707 | 0.9375 |
0.1257 | 49.0 | 1225 | 0.2281 | 0.95 |
0.1242 | 50.0 | 1250 | 0.2846 | 0.925 |
0.1169 | 51.0 | 1275 | 0.2766 | 0.9167 |
0.1701 | 52.0 | 1300 | 0.1858 | 0.9417 |
0.185 | 53.0 | 1325 | 0.3373 | 0.9292 |
0.0383 | 54.0 | 1350 | 0.3524 | 0.9208 |
0.0808 | 55.0 | 1375 | 0.3378 | 0.925 |
0.1444 | 56.0 | 1400 | 0.2609 | 0.9292 |
0.0798 | 57.0 | 1425 | 0.2635 | 0.9417 |
0.0324 | 58.0 | 1450 | 0.2550 | 0.9375 |
0.0669 | 59.0 | 1475 | 0.2466 | 0.9458 |
0.1389 | 60.0 | 1500 | 0.1992 | 0.95 |
0.0432 | 61.0 | 1525 | 0.2165 | 0.95 |
0.2076 | 62.0 | 1550 | 0.2718 | 0.9417 |
0.015 | 63.0 | 1575 | 0.2631 | 0.9458 |
0.0565 | 64.0 | 1600 | 0.2481 | 0.9417 |
0.0261 | 65.0 | 1625 | 0.2125 | 0.95 |
0.0136 | 66.0 | 1650 | 0.2464 | 0.95 |
0.0129 | 67.0 | 1675 | 0.2028 | 0.9542 |
0.1424 | 68.0 | 1700 | 0.1805 | 0.9542 |
0.0894 | 69.0 | 1725 | 0.2104 | 0.9458 |
0.04 | 70.0 | 1750 | 0.1842 | 0.9542 |
0.0424 | 71.0 | 1775 | 0.2014 | 0.9583 |
0.0364 | 72.0 | 1800 | 0.2029 | 0.9625 |
0.0112 | 73.0 | 1825 | 0.1695 | 0.9583 |
0.034 | 74.0 | 1850 | 0.1917 | 0.95 |
0.0253 | 75.0 | 1875 | 0.2004 | 0.9542 |
0.0089 | 76.0 | 1900 | 0.1981 | 0.9542 |
0.014 | 77.0 | 1925 | 0.1930 | 0.9542 |
0.0043 | 78.0 | 1950 | 0.1968 | 0.9417 |
0.0237 | 79.0 | 1975 | 0.1906 | 0.95 |
0.0031 | 80.0 | 2000 | 0.1904 | 0.95 |
Framework versions
- Transformers 4.21.1
- Pytorch 1.12.1
- Datasets 2.4.0
- Tokenizers 0.12.1