heartbeat-detection

This model is a fine-tuned version of ntu-spml/distilhubert on the audiofolder dataset. It achieves the following results on the evaluation set:

  • Loss: 0.6877
  • Accuracy: 1.0

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: 1e-06
  • train_batch_size: 8
  • eval_batch_size: 8
  • seed: 42
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.01
  • num_epochs: 100

Training results

Training Loss Epoch Step Validation Loss Accuracy
1.5999 1.0 8 1.5962 0.2222
1.5827 2.0 16 1.5749 0.8519
1.5617 3.0 24 1.5546 0.9630
1.541 4.0 32 1.5346 1.0
1.5215 5.0 40 1.5153 1.0
1.5014 6.0 48 1.4962 1.0
1.4817 7.0 56 1.4774 1.0
1.4622 8.0 64 1.4588 1.0
1.4444 9.0 72 1.4404 1.0
1.4244 10.0 80 1.4221 1.0
1.4068 11.0 88 1.4041 1.0
1.389 12.0 96 1.3862 1.0
1.3706 13.0 104 1.3685 1.0
1.3526 14.0 112 1.3510 1.0
1.3348 15.0 120 1.3337 1.0
1.3176 16.0 128 1.3165 1.0
1.3003 17.0 136 1.2997 1.0
1.283 18.0 144 1.2829 1.0
1.2667 19.0 152 1.2664 1.0
1.2508 20.0 160 1.2501 1.0
1.2345 21.0 168 1.2341 1.0
1.2189 22.0 176 1.2183 1.0
1.2052 23.0 184 1.2029 1.0
1.1882 24.0 192 1.1876 1.0
1.1734 25.0 200 1.1725 1.0
1.159 26.0 208 1.1579 1.0
1.1449 27.0 216 1.1436 1.0
1.1302 28.0 224 1.1295 1.0
1.1177 29.0 232 1.1156 1.0
1.1032 30.0 240 1.1021 1.0
1.0902 31.0 248 1.0887 1.0
1.0776 32.0 256 1.0758 1.0
1.0656 33.0 264 1.0628 1.0
1.0519 34.0 272 1.0502 1.0
1.0412 35.0 280 1.0378 1.0
1.0277 36.0 288 1.0255 1.0
1.0178 37.0 296 1.0135 1.0
1.0067 38.0 304 1.0021 1.0
0.9948 39.0 312 0.9907 1.0
0.9852 40.0 320 0.9799 1.0
0.973 41.0 328 0.9690 1.0
0.9635 42.0 336 0.9586 1.0
0.953 43.0 344 0.9484 1.0
0.943 44.0 352 0.9384 1.0
0.935 45.0 360 0.9284 1.0
0.9249 46.0 368 0.9188 1.0
0.916 47.0 376 0.9092 1.0
0.9068 48.0 384 0.9001 1.0
0.8989 49.0 392 0.8912 1.0
0.8904 50.0 400 0.8826 1.0
0.8831 51.0 408 0.8741 1.0
0.8744 52.0 416 0.8659 1.0
0.8676 53.0 424 0.8580 1.0
0.8592 54.0 432 0.8501 1.0
0.8531 55.0 440 0.8427 1.0
0.8445 56.0 448 0.8352 1.0
0.8382 57.0 456 0.8281 1.0
0.8308 58.0 464 0.8212 1.0
0.8264 59.0 472 0.8145 1.0
0.819 60.0 480 0.8079 1.0
0.8132 61.0 488 0.8016 1.0
0.8078 62.0 496 0.7954 1.0
0.801 63.0 504 0.7895 1.0
0.7971 64.0 512 0.7838 1.0
0.791 65.0 520 0.7783 1.0
0.7861 66.0 528 0.7729 1.0
0.7801 67.0 536 0.7677 1.0
0.7756 68.0 544 0.7627 1.0
0.7719 69.0 552 0.7579 1.0
0.7671 70.0 560 0.7533 1.0
0.762 71.0 568 0.7488 1.0
0.7581 72.0 576 0.7446 1.0
0.7538 73.0 584 0.7405 1.0
0.7512 74.0 592 0.7365 1.0
0.7467 75.0 600 0.7327 1.0
0.7435 76.0 608 0.7291 1.0
0.7412 77.0 616 0.7256 1.0
0.7368 78.0 624 0.7223 1.0
0.7345 79.0 632 0.7192 1.0
0.7313 80.0 640 0.7162 1.0
0.7285 81.0 648 0.7134 1.0
0.7261 82.0 656 0.7107 1.0
0.724 83.0 664 0.7082 1.0
0.7221 84.0 672 0.7058 1.0
0.7189 85.0 680 0.7036 1.0
0.7171 86.0 688 0.7016 1.0
0.7148 87.0 696 0.6997 1.0
0.7131 88.0 704 0.6979 1.0
0.7116 89.0 712 0.6963 1.0
0.7107 90.0 720 0.6948 1.0
0.7089 91.0 728 0.6934 1.0
0.7077 92.0 736 0.6922 1.0
0.7069 93.0 744 0.6911 1.0
0.7071 94.0 752 0.6902 1.0
0.705 95.0 760 0.6895 1.0
0.7045 96.0 768 0.6888 1.0
0.7038 97.0 776 0.6883 1.0
0.7029 98.0 784 0.6880 1.0
0.7041 99.0 792 0.6878 1.0
0.7029 100.0 800 0.6877 1.0

Framework versions

  • Transformers 4.39.3
  • Pytorch 2.2.1+cu121
  • Datasets 2.18.0
  • Tokenizers 0.15.2
Downloads last month
14
Safetensors
Model size
23.7M params
Tensor type
F32
·
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.

Model tree for Hemg/heartbeat-detection

Finetuned
(409)
this model

Evaluation results