<!-- 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. -->
kwav2vec-er-7-10-15000-long-16
This model is a fine-tuned version of facebook/wav2vec2-large-robust on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.3727
- Accuracy: 0.9
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: 0.0001
- train_batch_size: 16
- eval_batch_size: 16
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 1000
- num_epochs: 10
Training results
Training Loss | Epoch | Step | Validation Loss | Accuracy |
---|---|---|---|---|
1.943 | 0.06 | 300 | 1.9466 | 0.1429 |
1.8042 | 0.11 | 600 | 1.8149 | 0.2330 |
1.7755 | 0.17 | 900 | 1.6747 | 0.3358 |
1.619 | 0.23 | 1200 | 1.6355 | 0.3570 |
1.6172 | 0.29 | 1500 | 1.4936 | 0.4190 |
1.3796 | 0.34 | 1800 | 1.4567 | 0.4364 |
1.4774 | 0.4 | 2100 | 1.4279 | 0.4498 |
1.3554 | 0.46 | 2400 | 1.3803 | 0.4690 |
1.423 | 0.51 | 2700 | 1.2494 | 0.5330 |
1.2498 | 0.57 | 3000 | 1.3004 | 0.5117 |
1.3969 | 0.63 | 3300 | 1.2270 | 0.5363 |
1.3048 | 0.69 | 3600 | 1.2323 | 0.5409 |
1.1305 | 0.74 | 3900 | 1.3168 | 0.5164 |
1.0552 | 0.8 | 4200 | 1.1898 | 0.5543 |
1.2168 | 0.86 | 4500 | 1.2117 | 0.5482 |
1.0561 | 0.91 | 4800 | 1.0936 | 0.5896 |
1.2713 | 0.97 | 5100 | 1.0436 | 0.6053 |
1.0327 | 1.03 | 5400 | 1.0317 | 0.6178 |
1.2362 | 1.09 | 5700 | 1.0850 | 0.5945 |
1.0298 | 1.14 | 6000 | 1.0162 | 0.6187 |
1.0046 | 1.2 | 6300 | 0.9762 | 0.6367 |
1.0645 | 1.26 | 6600 | 1.0823 | 0.6046 |
1.1147 | 1.31 | 6900 | 0.9525 | 0.6497 |
0.883 | 1.37 | 7200 | 0.9250 | 0.6547 |
0.9322 | 1.43 | 7500 | 0.9857 | 0.642 |
1.1925 | 1.49 | 7800 | 0.9348 | 0.6582 |
1.0064 | 1.54 | 8100 | 0.8825 | 0.6687 |
0.9907 | 1.6 | 8400 | 0.8701 | 0.6831 |
0.8953 | 1.66 | 8700 | 0.7846 | 0.7108 |
1.0058 | 1.71 | 9000 | 0.8349 | 0.6950 |
0.7928 | 1.77 | 9300 | 0.8145 | 0.7018 |
0.8623 | 1.83 | 9600 | 0.8413 | 0.6890 |
0.991 | 1.89 | 9900 | 0.8114 | 0.6982 |
0.8623 | 1.94 | 10200 | 0.8128 | 0.7043 |
0.9766 | 2.0 | 10500 | 0.7279 | 0.7259 |
0.812 | 2.06 | 10800 | 0.7716 | 0.716 |
0.7853 | 2.11 | 11100 | 0.8257 | 0.7078 |
0.8994 | 2.17 | 11400 | 0.7944 | 0.7076 |
0.7399 | 2.23 | 11700 | 0.7252 | 0.7403 |
0.659 | 2.29 | 12000 | 0.7127 | 0.7432 |
0.6021 | 2.34 | 12300 | 0.7640 | 0.7297 |
0.8454 | 2.4 | 12600 | 0.6974 | 0.7518 |
0.781 | 2.46 | 12900 | 0.6417 | 0.7649 |
0.7437 | 2.51 | 13200 | 0.6468 | 0.7642 |
0.6632 | 2.57 | 13500 | 0.6097 | 0.7770 |
0.679 | 2.63 | 13800 | 0.6759 | 0.7552 |
0.5396 | 2.69 | 14100 | 0.6701 | 0.7609 |
0.6842 | 2.74 | 14400 | 0.6414 | 0.7662 |
0.4722 | 2.8 | 14700 | 0.6248 | 0.7711 |
0.7211 | 2.86 | 15000 | 0.6009 | 0.7770 |
0.8083 | 2.91 | 15300 | 0.6411 | 0.7674 |
0.6066 | 2.97 | 15600 | 0.6096 | 0.7751 |
0.6907 | 3.03 | 15900 | 0.6160 | 0.7793 |
0.6367 | 3.09 | 16200 | 0.6378 | 0.78 |
0.7089 | 3.14 | 16500 | 0.5958 | 0.7882 |
0.4632 | 3.2 | 16800 | 0.6068 | 0.7888 |
0.461 | 3.26 | 17100 | 0.6492 | 0.7738 |
0.5501 | 3.31 | 17400 | 0.5242 | 0.8083 |
0.4684 | 3.37 | 17700 | 0.5387 | 0.8059 |
0.4805 | 3.43 | 18000 | 0.6242 | 0.7794 |
0.5673 | 3.49 | 18300 | 0.5550 | 0.8034 |
0.5309 | 3.54 | 18600 | 0.5554 | 0.7977 |
0.5792 | 3.6 | 18900 | 0.5647 | 0.8014 |
0.3972 | 3.66 | 19200 | 0.5079 | 0.8215 |
0.4331 | 3.71 | 19500 | 0.5504 | 0.8039 |
0.5325 | 3.77 | 19800 | 0.6039 | 0.7895 |
0.4272 | 3.83 | 20100 | 0.5766 | 0.8016 |
0.4873 | 3.89 | 20400 | 0.5481 | 0.812 |
0.6392 | 3.94 | 20700 | 0.4955 | 0.8219 |
0.3925 | 4.0 | 21000 | 0.5238 | 0.8148 |
0.3812 | 4.06 | 21300 | 0.5694 | 0.8068 |
0.358 | 4.11 | 21600 | 0.5599 | 0.8110 |
0.5688 | 4.17 | 21900 | 0.4942 | 0.8234 |
0.503 | 4.23 | 22200 | 0.5018 | 0.8265 |
0.5035 | 4.29 | 22500 | 0.4893 | 0.8275 |
0.5212 | 4.34 | 22800 | 0.5099 | 0.8183 |
0.5498 | 4.4 | 23100 | 0.4601 | 0.832 |
0.4165 | 4.46 | 23400 | 0.5334 | 0.8190 |
0.4754 | 4.51 | 23700 | 0.5455 | 0.8172 |
0.4379 | 4.57 | 24000 | 0.5077 | 0.8290 |
0.545 | 4.63 | 24300 | 0.4609 | 0.8345 |
0.4391 | 4.69 | 24600 | 0.4454 | 0.8447 |
0.4662 | 4.74 | 24900 | 0.4633 | 0.8370 |
0.4708 | 4.8 | 25200 | 0.4698 | 0.8404 |
0.4163 | 4.86 | 25500 | 0.4124 | 0.8517 |
0.3001 | 4.91 | 25800 | 0.4435 | 0.8468 |
0.4455 | 4.97 | 26100 | 0.4376 | 0.8453 |
0.4108 | 5.03 | 26400 | 0.4455 | 0.8444 |
0.3197 | 5.09 | 26700 | 0.4366 | 0.8468 |
0.3493 | 5.14 | 27000 | 0.4547 | 0.8445 |
0.3584 | 5.2 | 27300 | 0.4369 | 0.8496 |
0.3741 | 5.26 | 27600 | 0.4193 | 0.8566 |
0.4225 | 5.31 | 27900 | 0.4167 | 0.8551 |
0.5255 | 5.37 | 28200 | 0.4253 | 0.8570 |
0.3775 | 5.43 | 28500 | 0.4649 | 0.8373 |
0.338 | 5.49 | 28800 | 0.3981 | 0.8589 |
0.374 | 5.54 | 29100 | 0.4789 | 0.84 |
0.4427 | 5.6 | 29400 | 0.4104 | 0.8607 |
0.3324 | 5.66 | 29700 | 0.4262 | 0.8539 |
0.3419 | 5.71 | 30000 | 0.4061 | 0.8587 |
0.3393 | 5.77 | 30300 | 0.4083 | 0.8617 |
0.4528 | 5.83 | 30600 | 0.4747 | 0.8417 |
0.5485 | 5.89 | 30900 | 0.4227 | 0.8588 |
0.3092 | 5.94 | 31200 | 0.4293 | 0.8627 |
0.3807 | 6.0 | 31500 | 0.3904 | 0.8649 |
0.2806 | 6.06 | 31800 | 0.4129 | 0.8654 |
0.2328 | 6.11 | 32100 | 0.4095 | 0.8674 |
0.315 | 6.17 | 32400 | 0.4271 | 0.8596 |
0.2961 | 6.23 | 32700 | 0.4059 | 0.8670 |
0.5028 | 6.29 | 33000 | 0.4555 | 0.8576 |
0.3411 | 6.34 | 33300 | 0.4113 | 0.8663 |
0.2555 | 6.4 | 33600 | 0.4282 | 0.8657 |
0.4199 | 6.46 | 33900 | 0.4286 | 0.8609 |
0.3403 | 6.51 | 34200 | 0.4193 | 0.8638 |
0.2486 | 6.57 | 34500 | 0.3921 | 0.8733 |
0.3294 | 6.63 | 34800 | 0.3810 | 0.8730 |
0.2884 | 6.69 | 35100 | 0.4108 | 0.8674 |
0.2757 | 6.74 | 35400 | 0.4006 | 0.8686 |
0.223 | 6.8 | 35700 | 0.4143 | 0.8670 |
0.2778 | 6.86 | 36000 | 0.3964 | 0.8729 |
0.2708 | 6.91 | 36300 | 0.4020 | 0.8717 |
0.4815 | 6.97 | 36600 | 0.4194 | 0.8697 |
0.2879 | 7.03 | 36900 | 0.3835 | 0.8752 |
0.1942 | 7.09 | 37200 | 0.4198 | 0.87 |
0.3093 | 7.14 | 37500 | 0.3683 | 0.8827 |
0.2402 | 7.2 | 37800 | 0.4087 | 0.8725 |
0.205 | 7.26 | 38100 | 0.3898 | 0.8794 |
0.2665 | 7.31 | 38400 | 0.3875 | 0.8796 |
0.361 | 7.37 | 38700 | 0.3753 | 0.8811 |
0.2704 | 7.43 | 39000 | 0.3973 | 0.8794 |
0.3198 | 7.49 | 39300 | 0.3690 | 0.8847 |
0.3003 | 7.54 | 39600 | 0.3829 | 0.8814 |
0.2854 | 7.6 | 39900 | 0.3917 | 0.8805 |
0.3197 | 7.66 | 40200 | 0.3969 | 0.8792 |
0.3 | 7.71 | 40500 | 0.4015 | 0.8817 |
0.2186 | 7.77 | 40800 | 0.3750 | 0.8850 |
0.3045 | 7.83 | 41100 | 0.4147 | 0.8796 |
0.2549 | 7.89 | 41400 | 0.3698 | 0.8872 |
0.21 | 7.94 | 41700 | 0.3672 | 0.8878 |
0.22 | 8.0 | 42000 | 0.3742 | 0.8876 |
0.2038 | 8.06 | 42300 | 0.3842 | 0.8850 |
0.1545 | 8.11 | 42600 | 0.3966 | 0.8836 |
0.2869 | 8.17 | 42900 | 0.4002 | 0.8850 |
0.2198 | 8.23 | 43200 | 0.3735 | 0.8893 |
0.2168 | 8.29 | 43500 | 0.3727 | 0.8918 |
0.1528 | 8.34 | 43800 | 0.4020 | 0.8869 |
0.2406 | 8.4 | 44100 | 0.3910 | 0.8870 |
0.1532 | 8.46 | 44400 | 0.3858 | 0.888 |
0.1937 | 8.51 | 44700 | 0.3906 | 0.8931 |
0.2456 | 8.57 | 45000 | 0.3868 | 0.8915 |
0.2585 | 8.63 | 45300 | 0.3871 | 0.8903 |
0.2257 | 8.69 | 45600 | 0.3774 | 0.8929 |
0.2344 | 8.74 | 45900 | 0.3815 | 0.8923 |
0.2175 | 8.8 | 46200 | 0.3833 | 0.8930 |
0.2785 | 8.86 | 46500 | 0.3960 | 0.8930 |
0.1932 | 8.91 | 46800 | 0.3820 | 0.8957 |
0.2476 | 8.97 | 47100 | 0.3755 | 0.8950 |
0.1389 | 9.03 | 47400 | 0.3884 | 0.8950 |
0.2625 | 9.09 | 47700 | 0.3846 | 0.8966 |
0.2484 | 9.14 | 48000 | 0.3728 | 0.8967 |
0.2221 | 9.2 | 48300 | 0.3787 | 0.8957 |
0.3583 | 9.26 | 48600 | 0.3816 | 0.8981 |
0.1621 | 9.31 | 48900 | 0.3906 | 0.8950 |
0.1917 | 9.37 | 49200 | 0.3818 | 0.8950 |
0.098 | 9.43 | 49500 | 0.3816 | 0.8970 |
0.2327 | 9.49 | 49800 | 0.3855 | 0.8975 |
0.196 | 9.54 | 50100 | 0.3847 | 0.8966 |
0.2066 | 9.6 | 50400 | 0.3738 | 0.8973 |
0.1835 | 9.66 | 50700 | 0.3719 | 0.8992 |
0.1939 | 9.71 | 51000 | 0.3819 | 0.8985 |
0.1882 | 9.77 | 51300 | 0.3711 | 0.8998 |
0.2127 | 9.83 | 51600 | 0.3801 | 0.8994 |
0.1589 | 9.89 | 51900 | 0.3737 | 0.9 |
0.139 | 9.94 | 52200 | 0.3734 | 0.8996 |
0.2246 | 10.0 | 52500 | 0.3727 | 0.9 |
Framework versions
- Transformers 4.30.2
- Pytorch 2.0.1+cu117
- Datasets 2.13.1
- Tokenizers 0.13.3