Hugging Face's logo

fydhfzh
/
hubert-classifier-aug

Audio Classification
Transformers
TensorBoard
Safetensors
hubert
Generated from Trainer
Inference Endpoints
Model card Files Metrics Community
hubert-classifier-aug / README.md
fydhfzh's picture
fydhfzh
End of training
8326048 verified
preview code
|
raw
history blame
9.56 kB
metadata
license: apache-2.0
base_model: facebook/hubert-base-ls960
tags:
  - generated_from_trainer
metrics:
  - accuracy
  - precision
  - recall
  - f1
model-index:
  - name: hubert-classifier-aug
    results: []

hubert-classifier-aug

This model is a fine-tuned version of facebook/hubert-base-ls960 on an unknown dataset. It achieves the following results on the evaluation set:

  • Loss: 0.4804
  • Accuracy: 0.8652
  • Precision: 0.8836
  • Recall: 0.8652
  • F1: 0.8626
  • Binary: 0.9049

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: 0.0001
  • train_batch_size: 32
  • eval_batch_size: 32
  • seed: 42
  • gradient_accumulation_steps: 4
  • total_train_batch_size: 128
  • optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
  • lr_scheduler_type: linear
  • lr_scheduler_warmup_steps: 500
  • num_epochs: 100
  • mixed_precision_training: Native AMP

Training results

Training Loss Epoch Step Validation Loss Accuracy Precision Recall F1 Binary
No log 0.19 50 3.9353 0.0539 0.0054 0.0539 0.0096 0.3232
No log 0.38 100 3.5299 0.0674 0.0168 0.0674 0.0214 0.3415
No log 0.58 150 3.2976 0.0809 0.0134 0.0809 0.0206 0.3550
No log 0.77 200 3.2171 0.0889 0.0148 0.0889 0.0244 0.3601
No log 0.96 250 3.0516 0.0970 0.0394 0.0970 0.0425 0.3674
No log 1.15 300 2.9416 0.1105 0.0476 0.1105 0.0455 0.3765
No log 1.34 350 2.8641 0.1348 0.0564 0.1348 0.0601 0.3911
No log 1.53 400 2.6831 0.2129 0.1140 0.2129 0.1256 0.4474
No log 1.73 450 2.4972 0.2534 0.1646 0.2534 0.1750 0.4757
3.3455 1.92 500 2.3968 0.2911 0.2066 0.2911 0.2067 0.4957
3.3455 2.11 550 2.2022 0.3720 0.2673 0.3720 0.2752 0.5606
3.3455 2.3 600 1.9843 0.4394 0.3474 0.4394 0.3584 0.6067
3.3455 2.49 650 1.8981 0.4582 0.3704 0.4582 0.3817 0.6175
3.3455 2.68 700 1.7742 0.4987 0.4631 0.4987 0.4378 0.6474
3.3455 2.88 750 1.5820 0.5337 0.4673 0.5337 0.4661 0.6749
3.3455 3.07 800 1.5259 0.5795 0.5610 0.5795 0.5335 0.7024
3.3455 3.26 850 1.3476 0.6361 0.5896 0.6361 0.5854 0.7477
3.3455 3.45 900 1.2645 0.6253 0.5957 0.6253 0.5700 0.7372
3.3455 3.64 950 1.2000 0.6577 0.6802 0.6577 0.6299 0.7598
2.2111 3.84 1000 1.1048 0.7170 0.7165 0.7170 0.6874 0.8003
2.2111 4.03 1050 1.0875 0.6900 0.6668 0.6900 0.6479 0.7798
2.2111 4.22 1100 1.0306 0.7197 0.7434 0.7197 0.6975 0.8040
2.2111 4.41 1150 0.9579 0.7520 0.7813 0.7520 0.7326 0.8261
2.2111 4.6 1200 0.8882 0.7358 0.7268 0.7358 0.7068 0.8143
2.2111 4.79 1250 0.9165 0.7574 0.7754 0.7574 0.7388 0.8286
2.2111 4.99 1300 0.8355 0.7763 0.7844 0.7763 0.7553 0.8447
2.2111 5.18 1350 0.7983 0.7871 0.8111 0.7871 0.7784 0.8523
2.2111 5.37 1400 0.7626 0.7978 0.8167 0.7978 0.7841 0.8569
2.2111 5.56 1450 0.7248 0.7898 0.7903 0.7898 0.7745 0.8553
1.6095 5.75 1500 0.7431 0.8005 0.8391 0.8005 0.7914 0.8598
1.6095 5.94 1550 0.7692 0.7790 0.8071 0.7790 0.7662 0.8466
1.6095 6.14 1600 0.6135 0.8302 0.8556 0.8302 0.8229 0.8814
1.6095 6.33 1650 0.6347 0.8221 0.8377 0.8221 0.8156 0.8757
1.6095 6.52 1700 0.6184 0.8221 0.8529 0.8221 0.8184 0.8747
1.6095 6.71 1750 0.6224 0.8221 0.8540 0.8221 0.8155 0.8757
1.6095 6.9 1800 0.6251 0.8194 0.8267 0.8194 0.8087 0.8728
1.6095 7.09 1850 0.5821 0.8383 0.8597 0.8383 0.8346 0.8860
1.6095 7.29 1900 0.6197 0.8059 0.8438 0.8059 0.8040 0.8644
1.6095 7.48 1950 0.5886 0.8275 0.8640 0.8275 0.8269 0.8784
1.3113 7.67 2000 0.5720 0.8410 0.8664 0.8410 0.8397 0.8889
1.3113 7.86 2050 0.6286 0.8248 0.8530 0.8248 0.8168 0.8765
1.3113 8.05 2100 0.5317 0.8329 0.8494 0.8329 0.8240 0.8822
1.3113 8.25 2150 0.4692 0.8571 0.8795 0.8571 0.8521 0.9003
1.3113 8.44 2200 0.5233 0.8598 0.8786 0.8598 0.8576 0.9011
1.3113 8.63 2250 0.5538 0.8329 0.8454 0.8329 0.8262 0.8822
1.3113 8.82 2300 0.5662 0.8167 0.8430 0.8167 0.8103 0.8712
1.3113 9.01 2350 0.5567 0.8491 0.8717 0.8491 0.8443 0.8946
1.3113 9.2 2400 0.5064 0.8464 0.8713 0.8464 0.8415 0.8916
1.3113 9.4 2450 0.5497 0.8544 0.8697 0.8544 0.8461 0.8973
1.1269 9.59 2500 0.5250 0.8544 0.8769 0.8544 0.8467 0.8981
1.1269 9.78 2550 0.5301 0.8571 0.8776 0.8571 0.8552 0.9003
1.1269 9.97 2600 0.4692 0.8733 0.8979 0.8733 0.8711 0.9105
1.1269 10.16 2650 0.5143 0.8544 0.8738 0.8544 0.8512 0.8965
1.1269 10.35 2700 0.5419 0.8383 0.8603 0.8383 0.8372 0.8860
1.1269 10.55 2750 0.6064 0.8275 0.8610 0.8275 0.8251 0.8784
1.1269 10.74 2800 0.4815 0.8760 0.9038 0.8760 0.8749 0.9124
1.1269 10.93 2850 0.4908 0.8652 0.8907 0.8652 0.8611 0.9059
1.1269 11.12 2900 0.5417 0.8491 0.8773 0.8491 0.8468 0.8935
1.1269 11.31 2950 0.5086 0.8518 0.8698 0.8518 0.8493 0.8954
1.0103 11.51 3000 0.5147 0.8518 0.8660 0.8518 0.8505 0.8954
1.0103 11.7 3050 0.5247 0.8571 0.8838 0.8571 0.8564 0.9000
1.0103 11.89 3100 0.4810 0.8625 0.8837 0.8625 0.8591 0.9040
1.0103 12.08 3150 0.4950 0.8706 0.8990 0.8706 0.8686 0.9097
1.0103 12.27 3200 0.4966 0.8544 0.8718 0.8544 0.8501 0.8965
1.0103 12.46 3250 0.4522 0.8598 0.8759 0.8598 0.8561 0.9030
1.0103 12.66 3300 0.5552 0.8437 0.8677 0.8437 0.8401 0.8916
1.0103 12.85 3350 0.5489 0.8248 0.8556 0.8248 0.8247 0.8776
1.0103 13.04 3400 0.5635 0.8598 0.8827 0.8598 0.8594 0.9011
1.0103 13.23 3450 0.5023 0.8652 0.8862 0.8652 0.8634 0.9067
0.912 13.42 3500 0.4804 0.8652 0.8836 0.8652 0.8626 0.9049
0.912 13.61 3550 0.4868 0.8598 0.8849 0.8598 0.8575 0.9013
0.912 13.81 3600 0.5493 0.8571 0.8852 0.8571 0.8533 0.8992
0.912 14.0 3650 0.5699 0.8437 0.8659 0.8437 0.8418 0.8908
0.912 14.19 3700 0.5606 0.8437 0.8692 0.8437 0.8395 0.8908
0.912 14.38 3750 0.5685 0.8491 0.8730 0.8491 0.8458 0.8935
0.912 14.57 3800 0.5088 0.8625 0.8792 0.8625 0.8606 0.9030
0.912 14.77 3850 0.5566 0.8437 0.8729 0.8437 0.8367 0.8898

Framework versions

  • Transformers 4.38.2
  • Pytorch 2.3.0
  • Datasets 2.19.1
  • Tokenizers 0.15.1