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

This model is a fine-tuned version of gary109/ai-light-dance_drums_ft_pretrain_wav2vec2-base-new_onset-idmt-2 on the GARY109/AI_LIGHT_DANCE - ONSET-IDMT-2 dataset. It achieves the following results on the evaluation set:

  • Loss: 0.2794
  • Wer: 0.2733

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 0.3089 0.2856
0.2871 2.0 18 0.3208 0.28
0.2997 3.0 27 0.3948 0.2878
0.299 4.0 36 0.3137 0.3011
0.3462 5.0 45 0.3067 0.2689
0.3098 6.0 54 0.3271 0.2811
0.2812 7.0 63 0.4907 0.26
0.3151 8.0 72 0.5852 0.2778
0.3038 9.0 81 0.2981 0.2767
0.3248 10.0 90 0.3129 0.2811
0.3248 11.0 99 0.4090 0.2767
0.3106 12.0 108 0.5354 0.3
0.2702 13.0 117 0.5543 0.3
0.3021 14.0 126 0.5437 0.2689
0.2622 15.0 135 0.5898 0.2778
0.2465 16.0 144 0.2900 0.2722
0.3077 17.0 153 0.4407 0.2544
0.2959 18.0 162 0.4079 0.2944
0.2843 19.0 171 0.5042 0.2722
0.254 20.0 180 0.3851 0.2878
0.254 21.0 189 0.3912 0.2678
0.2532 22.0 198 0.4699 0.2578
0.3011 23.0 207 0.7466 0.2744
0.2601 24.0 216 0.4238 0.28
0.2873 25.0 225 0.3817 0.2456
0.2791 26.0 234 0.3488 0.2489
0.2399 27.0 243 0.2980 0.2611
0.2592 28.0 252 0.2942 0.27
0.2191 29.0 261 0.2921 0.2833
0.2285 30.0 270 0.2851 0.2744
0.2285 31.0 279 0.2794 0.2733
0.2489 32.0 288 0.3036 0.2678
0.2445 33.0 297 0.2851 0.2678
0.2261 34.0 306 0.2864 0.2733
0.2391 35.0 315 0.3055 0.2611
0.3939 36.0 324 0.2927 0.26
0.2521 37.0 333 0.3470 0.2578
0.2378 38.0 342 0.2841 0.2656
0.2653 39.0 351 0.2889 0.2389
0.2235 40.0 360 0.3176 0.25
0.2235 41.0 369 0.3188 0.2667
0.2474 42.0 378 0.3782 0.2633
0.222 43.0 387 0.3201 0.2767
0.2411 44.0 396 0.3416 0.2722
0.2561 45.0 405 0.3050 0.2711
0.2169 46.0 414 0.3968 0.2511
0.2296 47.0 423 0.3721 0.2567
0.1989 48.0 432 0.3205 0.2667
0.2408 49.0 441 0.4524 0.2489
0.2163 50.0 450 0.4850 0.2567
0.2163 51.0 459 0.3777 0.2711
0.2001 52.0 468 0.5526 0.2644
0.2373 53.0 477 0.5141 0.2589
0.2132 54.0 486 0.5408 0.2611
0.2687 55.0 495 0.5389 0.2678
0.2244 56.0 504 0.5729 0.2578
0.2102 57.0 513 0.6249 0.2489
0.2076 58.0 522 0.5538 0.25
0.208 59.0 531 0.5499 0.2467
0.2167 60.0 540 0.6481 0.2433
0.2167 61.0 549 0.6797 0.2589
0.2218 62.0 558 0.5401 0.2656
0.2102 63.0 567 0.5152 0.26
0.2176 64.0 576 0.5581 0.26
0.2068 65.0 585 0.7225 0.2533
0.2123 66.0 594 0.6330 0.2633
0.2212 67.0 603 0.5943 0.2589
0.2013 68.0 612 0.7557 0.25
0.2304 69.0 621 0.9144 0.2467
0.209 70.0 630 0.7790 0.24
0.209 71.0 639 0.6203 0.2411
0.191 72.0 648 0.6280 0.2322
0.2313 73.0 657 0.5491 0.2378
0.1869 74.0 666 0.4653 0.2411
0.2313 75.0 675 0.6016 0.2489
0.1806 76.0 684 0.6492 0.2478
0.1934 77.0 693 0.6185 0.2478
0.1954 78.0 702 0.5618 0.2489
0.2077 79.0 711 0.5760 0.2522
0.2052 80.0 720 0.6172 0.25
0.2052 81.0 729 0.6859 0.2467
0.1804 82.0 738 0.7643 0.2422
0.1995 83.0 747 0.8360 0.2367
0.1869 84.0 756 0.6984 0.2489
0.2135 85.0 765 0.6759 0.2422
0.178 86.0 774 0.6791 0.2444
0.1734 87.0 783 0.7284 0.2411
0.1881 88.0 792 0.8172 0.2344
0.1625 89.0 801 0.8061 0.2356
0.181 90.0 810 0.7644 0.2389
0.181 91.0 819 0.7413 0.24
0.1942 92.0 828 0.6439 0.2433
0.1806 93.0 837 0.6250 0.2467
0.1651 94.0 846 0.6517 0.2433
0.1833 95.0 855 0.6628 0.2389
0.1873 96.0 864 0.6582 0.2378
0.1672 97.0 873 0.6548 0.2389
0.1871 98.0 882 0.6655 0.24
0.2429 99.0 891 0.6695 0.24
0.1832 100.0 900 0.6700 0.2389

Framework versions

  • Transformers 4.25.0.dev0
  • Pytorch 1.8.1+cu111
  • Datasets 2.7.1.dev0
  • Tokenizers 0.13.2
Downloads last month
14
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.