Edit model card

whisper-tiny-south-indic-audio-abuse-feature

This model is a fine-tuned version of parambharat/whisper-tiny-south-indic on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 0.8754
  • Accuracy: 0.8091
  • Macro F1-score: 0.7326

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-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: 12

Training results

Training Loss Epoch Step Validation Loss Accuracy Macro F1-score
7.6265 0.29 10 7.5999 0.0 0.0
7.5514 0.58 20 7.4739 0.0 0.0
7.3762 0.86 30 7.2108 0.7073 0.2762
7.0242 1.15 40 6.6877 0.7114 0.4157
6.421 1.44 50 6.0112 0.7114 0.4157
5.7786 1.73 60 5.4468 0.7114 0.4157
5.2429 2.01 70 4.9849 0.7114 0.4157
4.8449 2.3 80 4.5921 0.7114 0.4157
4.5013 2.59 90 4.2484 0.7114 0.4157
4.1317 2.88 100 3.9375 0.7114 0.4157
3.8904 3.17 110 3.6543 0.7114 0.4157
3.5933 3.45 120 3.3909 0.7114 0.4157
3.3129 3.74 130 3.1460 0.7114 0.4157
3.0954 4.03 140 2.9201 0.7114 0.4157
2.8817 4.32 150 2.7098 0.7114 0.4157
2.7003 4.6 160 2.5173 0.7114 0.4157
2.5074 4.89 170 2.3380 0.7114 0.4157
2.3684 5.18 180 2.1744 0.7114 0.4157
2.1876 5.47 190 2.0227 0.7114 0.4157
2.0526 5.76 200 1.8856 0.7114 0.4157
1.8551 6.04 210 1.7647 0.7114 0.4157
1.7855 6.33 220 1.6472 0.7114 0.4157
1.7239 6.62 230 1.5505 0.7358 0.4996
1.5197 6.91 240 1.4623 0.7114 0.4157
1.485 7.19 250 1.3723 0.7439 0.5250
1.4318 7.48 260 1.2978 0.7642 0.5914
1.3828 7.77 270 1.2308 0.7805 0.6390
1.1559 8.06 280 1.1848 0.7398 0.5124
1.1692 8.35 290 1.1189 0.7886 0.6581
1.2058 8.63 300 1.0709 0.7846 0.6486
1.1676 8.92 310 1.0313 0.7886 0.6635
1.0257 9.21 320 0.9966 0.7886 0.6581
0.9943 9.5 330 0.9652 0.7886 0.6635
0.9894 9.78 340 0.9398 0.7927 0.6726
1.0581 10.07 350 0.9206 0.8171 0.7313
0.9395 10.36 360 0.9003 0.8008 0.6992
0.9766 10.65 370 0.8873 0.7886 0.6687
0.9291 10.94 380 0.8759 0.8089 0.7155
0.8967 11.22 390 0.8704 0.7927 0.6726
0.9463 11.51 400 0.8656 0.7967 0.6862

Framework versions

  • Transformers 4.33.0
  • Pytorch 2.0.0
  • Datasets 2.1.0
  • Tokenizers 0.13.3
Downloads last month
0
Inference API
or
This model can be loaded on Inference API (serverless).

Finetuned from