audioclass-alpha

This model is a fine-tuned version of facebook/wav2vec2-base-960h on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.0963
  • Accuracy: 0.9819

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: 3e-05
  • train_batch_size: 16
  • eval_batch_size: 16
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 64
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_ratio: 0.1
  • num_epochs: 50
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy
3.3678 1.0 62 3.3612 0.0408
3.349 2.0 124 3.3402 0.0385
3.2935 3.0 186 3.2047 0.2494
2.9643 4.0 248 2.7250 0.5102
2.4158 5.0 310 2.1914 0.6621
1.9634 6.0 372 1.7440 0.7800
1.6144 7.0 434 1.3680 0.8503
1.2939 8.0 496 1.0948 0.8390
1.0933 9.0 558 0.8783 0.8776
0.8596 10.0 620 0.7053 0.9048
0.6664 11.0 682 0.6020 0.9184
0.5843 12.0 744 0.5392 0.9048
0.5714 13.0 806 0.4380 0.9297
0.4395 14.0 868 0.4434 0.9252
0.323 15.0 930 0.3000 0.9524
0.3218 16.0 992 0.2418 0.9546
0.3026 17.0 1054 0.2462 0.9524
0.2531 18.0 1116 0.2003 0.9660
0.2702 19.0 1178 0.1883 0.9637
0.2368 20.0 1240 0.1612 0.9728
0.2121 21.0 1302 0.1981 0.9637
0.2011 22.0 1364 0.1635 0.9683
0.1875 23.0 1426 0.1454 0.9728
0.1415 24.0 1488 0.1433 0.9683
0.1162 25.0 1550 0.1504 0.9660
0.0946 26.0 1612 0.1759 0.9615
0.1032 27.0 1674 0.1206 0.9751
0.095 28.0 1736 0.1123 0.9773
0.1526 29.0 1798 0.1267 0.9728
0.1003 30.0 1860 0.0953 0.9796
0.1371 31.0 1922 0.1158 0.9751
0.0765 32.0 1984 0.0963 0.9819
0.1152 33.0 2046 0.0929 0.9819
0.1344 34.0 2108 0.1103 0.9796
0.1067 35.0 2170 0.1065 0.9773
0.0847 36.0 2232 0.0898 0.9819
0.0835 37.0 2294 0.0934 0.9819
0.1009 38.0 2356 0.1136 0.9796
0.1272 39.0 2418 0.1315 0.9751
0.0463 40.0 2480 0.1127 0.9796
0.085 41.0 2542 0.0985 0.9796
0.0431 42.0 2604 0.0964 0.9773
0.0698 43.0 2666 0.1128 0.9773
0.0493 44.0 2728 0.0934 0.9796
0.1208 45.0 2790 0.0882 0.9819
0.0536 46.0 2852 0.0932 0.9796
0.064 47.0 2914 0.1008 0.9796
0.0538 48.0 2976 0.1094 0.9796
0.0774 49.0 3038 0.1081 0.9796
0.0379 50.0 3100 0.1085 0.9796

Framework versions

  • Transformers 4.36.0.dev0
  • Pytorch 2.1.0+cu118
  • Datasets 2.14.6
  • Tokenizers 0.14.1
Downloads last month
18
Safetensors
Model size
94.6M 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 giovaldir/audioclass-alpha

Finetuned
(122)
this model