lyrics-generator / README.md
qiny17's picture
update model card README.md
7d54a5b
metadata
license: mit
tags:
  - generated_from_trainer
model-index:
  - name: lyrics-generator-v2
    results: []

lyrics-generator-v2

This model is a fine-tuned version of gpt2 on the None dataset. It achieves the following results on the evaluation set:

  • Loss: 2.1114

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: 5e-05
  • 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
  • num_epochs: 10

Training results

Training Loss Epoch Step Validation Loss
3.4325 0.36 200 2.3267
2.4121 0.71 400 2.2693
2.3582 1.07 600 2.2401
2.2903 1.42 800 2.2182
2.2738 1.78 1000 2.2046
2.2322 2.14 1200 2.1922
2.1933 2.49 1400 2.1832
2.1944 2.85 1600 2.1736
2.1632 3.2 1800 2.1648
2.1366 3.56 2000 2.1554
2.1492 3.91 2200 2.1491
2.1108 4.27 2400 2.1472
2.0882 4.63 2600 2.1422
2.0971 4.98 2800 2.1343
2.0829 5.34 3000 2.1318
2.042 5.69 3200 2.1280
2.0375 6.05 3400 2.1261
2.0146 6.41 3600 2.1245
2.0551 6.76 3800 2.1217
1.992 7.12 4000 2.1182
1.9994 7.47 4200 2.1170
2.0189 7.83 4400 2.1156
1.9795 8.19 4600 2.1133
2.0101 8.54 4800 2.1143
1.9864 8.9 5000 2.1111
1.9602 9.25 5200 2.1120
1.9899 9.61 5400 2.1117
1.9928 9.96 5600 2.1114

Framework versions

  • Transformers 4.25.1
  • Pytorch 1.12.1+cu113
  • Datasets 2.7.1
  • Tokenizers 0.13.2