<!-- 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_one_speaker_all_classes_TTS
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: 0.2062
- Accuracy: 0.9579
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: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 128
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 60
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|---|---|---|---|
| 3.6865 | 0.99 | 34 | 3.6873 | 0.025 |
| 3.6155 | 1.99 | 68 | 3.4150 | 0.2188 |
| 2.6933 | 2.99 | 102 | 2.4527 | 0.4625 |
| 1.789 | 3.99 | 136 | 1.5249 | 0.7246 |
| 0.8812 | 4.99 | 170 | 0.7804 | 0.8708 |
| 0.4054 | 5.99 | 204 | 0.6304 | 0.8558 |
| 0.3481 | 6.99 | 238 | 0.5552 | 0.8667 |
| 0.238 | 7.99 | 272 | 0.4142 | 0.9113 |
| 0.1981 | 8.99 | 306 | 0.3007 | 0.9354 |
| 0.1254 | 9.99 | 340 | 0.2556 | 0.9479 |
| 0.1356 | 10.99 | 374 | 0.5148 | 0.8825 |
| 0.1263 | 11.99 | 408 | 0.3228 | 0.9308 |
| 0.1074 | 12.99 | 442 | 0.3085 | 0.9279 |
| 0.0756 | 13.99 | 476 | 0.4546 | 0.9029 |
| 0.0763 | 14.99 | 510 | 0.4045 | 0.9133 |
| 0.0902 | 15.99 | 544 | 0.3123 | 0.9287 |
| 0.1134 | 16.99 | 578 | 0.2054 | 0.9504 |
| 0.0943 | 17.99 | 612 | 0.2871 | 0.93 |
| 0.0511 | 18.99 | 646 | 0.3628 | 0.9292 |
| 0.0525 | 19.99 | 680 | 0.2228 | 0.9471 |
| 0.0769 | 20.99 | 714 | 0.3069 | 0.9329 |
| 0.0564 | 21.99 | 748 | 0.2658 | 0.9358 |
| 0.0319 | 22.99 | 782 | 0.2886 | 0.9387 |
| 0.0485 | 23.99 | 816 | 0.2342 | 0.9467 |
| 0.0542 | 24.99 | 850 | 0.3723 | 0.9287 |
| 0.0478 | 25.99 | 884 | 0.2890 | 0.9396 |
| 0.0373 | 26.99 | 918 | 0.2849 | 0.9383 |
| 0.0437 | 27.99 | 952 | 0.3886 | 0.9237 |
| 0.02 | 28.99 | 986 | 0.2672 | 0.9387 |
| 0.0379 | 29.99 | 1020 | 0.2946 | 0.9363 |
| 0.0253 | 30.99 | 1054 | 0.2499 | 0.9433 |
| 0.0256 | 31.99 | 1088 | 0.2967 | 0.9337 |
| 0.029 | 32.99 | 1122 | 0.2577 | 0.9458 |
| 0.0427 | 33.99 | 1156 | 0.2899 | 0.9396 |
| 0.0167 | 34.99 | 1190 | 0.2984 | 0.9437 |
| 0.0334 | 35.99 | 1224 | 0.4822 | 0.9175 |
| 0.0288 | 36.99 | 1258 | 0.2802 | 0.9417 |
| 0.017 | 37.99 | 1292 | 0.2233 | 0.9504 |
| 0.0064 | 38.99 | 1326 | 0.2657 | 0.9429 |
| 0.0176 | 39.99 | 1360 | 0.2062 | 0.9579 |
| 0.0307 | 40.99 | 1394 | 0.3633 | 0.9275 |
| 0.0208 | 41.99 | 1428 | 0.3059 | 0.9421 |
| 0.0091 | 42.99 | 1462 | 0.2488 | 0.9483 |
| 0.0121 | 43.99 | 1496 | 0.2397 | 0.9496 |
| 0.0106 | 44.99 | 1530 | 0.2958 | 0.9413 |
| 0.0176 | 45.99 | 1564 | 0.2243 | 0.9525 |
| 0.0153 | 46.99 | 1598 | 0.2293 | 0.9537 |
| 0.011 | 47.99 | 1632 | 0.2654 | 0.9496 |
| 0.0237 | 48.99 | 1666 | 0.2252 | 0.9533 |
| 0.0053 | 49.99 | 1700 | 0.2380 | 0.9483 |
| 0.0142 | 50.99 | 1734 | 0.2590 | 0.9467 |
| 0.0259 | 51.99 | 1768 | 0.2363 | 0.9508 |
| 0.0062 | 52.99 | 1802 | 0.2451 | 0.9496 |
| 0.0123 | 53.99 | 1836 | 0.2546 | 0.9479 |
| 0.011 | 54.99 | 1870 | 0.2578 | 0.9487 |
| 0.0143 | 55.99 | 1904 | 0.2770 | 0.945 |
| 0.015 | 56.99 | 1938 | 0.2869 | 0.9421 |
| 0.0099 | 57.99 | 1972 | 0.2922 | 0.9429 |
| 0.0086 | 58.99 | 2006 | 0.2783 | 0.9437 |
| 0.013 | 59.99 | 2040 | 0.2748 | 0.9433 |
Framework versions
- Transformers 4.21.1
- Pytorch 1.12.1
- Datasets 2.4.0
- Tokenizers 0.12.1
