Edit model card

ai-light-dance_drums_ft_pretrain_wav2vec2-base-new_onset-idmt-2_8k

This model is a fine-tuned version of gary109/ai-light-dance_drums_pretrain_wav2vec2-base-new on the GARY109/AI_LIGHT_DANCE - ONSET-IDMT-2 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.5029
  • Wer: 0.3178

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_steps: 30
  • num_epochs: 100.0
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Wer
No log 1.0 9 101.8046 0.98
17.4958 2.0 18 82.4920 1.0
16.2087 3.0 27 36.1388 1.0
6.2942 4.0 36 8.3267 1.0
2.0411 5.0 45 6.8215 1.0
1.554 6.0 54 5.3847 1.0
1.6215 7.0 63 4.4645 1.0
1.4962 8.0 72 3.2211 1.0
1.3825 9.0 81 2.5513 1.0
1.3443 10.0 90 2.8582 1.0
1.3443 11.0 99 2.5446 1.0
1.3096 12.0 108 2.0211 0.9956
1.3361 13.0 117 1.8110 0.9944
1.2862 14.0 126 1.7796 0.9933
1.2556 15.0 135 1.7301 0.9922
1.1959 16.0 144 1.4245 0.9989
1.1161 17.0 153 1.1932 0.5678
0.8853 18.0 162 1.2726 0.4922
0.7996 19.0 171 1.0841 0.5511
0.8165 20.0 180 1.4062 0.4411
0.8165 21.0 189 1.4219 0.3367
0.6807 22.0 198 1.2107 0.3344
0.7315 23.0 207 1.1420 0.3189
0.6203 24.0 216 1.0770 0.3778
0.6552 25.0 225 1.1095 0.3789
0.5618 26.0 234 1.0004 0.3478
0.5311 27.0 243 0.8811 0.3311
0.5391 28.0 252 0.8163 0.3678
0.5275 29.0 261 1.0000 0.3311
0.4965 30.0 270 0.7320 0.37
0.4965 31.0 279 0.9643 0.3389
0.4909 32.0 288 0.7663 0.3589
0.5218 33.0 297 0.9004 0.3489
0.4991 34.0 306 0.7342 0.38
0.4883 35.0 315 0.7959 0.3389
0.4902 36.0 324 0.6892 0.3378
0.4447 37.0 333 0.6480 0.3333
0.4458 38.0 342 0.6198 0.3333
0.4607 39.0 351 0.6081 0.3111
0.4352 40.0 360 0.6748 0.3156
0.4352 41.0 369 0.6885 0.3256
0.4286 42.0 378 0.6806 0.3333
0.4314 43.0 387 0.7855 0.3222
0.4476 44.0 396 0.6569 0.3144
0.4815 45.0 405 0.5389 0.3033
0.36 46.0 414 0.5550 0.3011
0.4516 47.0 423 0.5924 0.3144
0.3682 48.0 432 0.7275 0.3056
0.4371 49.0 441 0.7051 0.3089
0.4004 50.0 450 0.5669 0.3078
0.4004 51.0 459 0.5029 0.3178
0.3298 52.0 468 0.6150 0.32
0.4083 53.0 477 0.5882 0.33
0.4022 54.0 486 0.7253 0.3144
0.4465 55.0 495 0.6808 0.3111
0.3955 56.0 504 0.6002 0.3133
0.3877 57.0 513 0.7593 0.3056
0.3486 58.0 522 0.6764 0.3189
0.3782 59.0 531 0.6772 0.3133
0.3599 60.0 540 0.8846 0.3111
0.3599 61.0 549 0.9458 0.3233
0.3424 62.0 558 0.8399 0.3233
0.3652 63.0 567 0.8266 0.3133
0.3327 64.0 576 0.7813 0.3078
0.3603 65.0 585 0.8066 0.3156
0.3401 66.0 594 0.7960 0.3067
0.3797 67.0 603 0.8513 0.2989
0.3353 68.0 612 0.8319 0.2722
0.3909 69.0 621 0.8244 0.2878
0.3263 70.0 630 0.9539 0.3022
0.3263 71.0 639 1.0030 0.2922
0.3102 72.0 648 0.9875 0.3044
0.3577 73.0 657 0.9030 0.2978
0.2953 74.0 666 0.9392 0.2889
0.3644 75.0 675 0.9089 0.2878
0.3231 76.0 684 0.9264 0.2844
0.3078 77.0 693 1.0536 0.2911
0.4503 78.0 702 0.9473 0.2967
0.3492 79.0 711 0.8909 0.3089
0.347 80.0 720 0.8532 0.3067
0.347 81.0 729 0.9553 0.2833
0.2949 82.0 738 1.0111 0.2867
0.3447 83.0 747 0.9160 0.3011
0.2878 84.0 756 0.8401 0.2989
0.3229 85.0 765 0.8815 0.2911
0.276 86.0 774 0.8802 0.2911
0.3469 87.0 783 0.9121 0.29
0.3044 88.0 792 0.8934 0.2933
0.2885 89.0 801 0.8806 0.2967
0.3365 90.0 810 0.9037 0.2844
0.3365 91.0 819 0.9218 0.2867
0.3239 92.0 828 0.9228 0.2844
0.3219 93.0 837 0.9167 0.2844
0.2736 94.0 846 0.9495 0.2878
0.3587 95.0 855 0.9997 0.2844
0.3386 96.0 864 0.9977 0.2856
0.2895 97.0 873 0.9964 0.2889
0.3496 98.0 882 0.9765 0.2889
0.2789 99.0 891 0.9713 0.2878
0.3284 100.0 900 0.9687 0.2889

Framework versions

  • Transformers 4.25.0.dev0
  • Pytorch 1.8.1+cu111
  • Datasets 2.7.1.dev0
  • Tokenizers 0.13.2
Downloads last month
2