<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
gpt2_left_out_aochildes
This model is a fine-tuned version of gpt2 on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 3.9775
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: 0.0005
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 1000
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
6.0187 | 0.25 | 500 | 5.1088 |
4.7853 | 0.5 | 1000 | 4.7258 |
4.4807 | 0.75 | 1500 | 4.4907 |
4.2957 | 0.99 | 2000 | 4.3381 |
4.074 | 1.24 | 2500 | 4.2557 |
4.0119 | 1.49 | 3000 | 4.1762 |
3.9329 | 1.74 | 3500 | 4.1028 |
3.8756 | 1.99 | 4000 | 4.0331 |
3.6686 | 2.24 | 4500 | 4.0161 |
3.6659 | 2.49 | 5000 | 3.9698 |
3.6573 | 2.73 | 5500 | 3.9297 |
3.6216 | 2.98 | 6000 | 3.8913 |
3.4206 | 3.23 | 6500 | 3.9039 |
3.4207 | 3.48 | 7000 | 3.8831 |
3.4302 | 3.73 | 7500 | 3.8568 |
3.4251 | 3.98 | 8000 | 3.8265 |
3.1989 | 4.23 | 8500 | 3.8662 |
3.2115 | 4.48 | 9000 | 3.8490 |
3.2284 | 4.72 | 9500 | 3.8314 |
3.2306 | 4.97 | 10000 | 3.8143 |
2.985 | 5.22 | 10500 | 3.8640 |
3.0071 | 5.47 | 11000 | 3.8580 |
3.0158 | 5.72 | 11500 | 3.8439 |
3.0238 | 5.97 | 12000 | 3.8284 |
2.7909 | 6.22 | 12500 | 3.8847 |
2.7777 | 6.46 | 13000 | 3.8899 |
2.801 | 6.71 | 13500 | 3.8799 |
2.8094 | 6.96 | 14000 | 3.8719 |
2.6007 | 7.21 | 14500 | 3.9203 |
2.5795 | 7.46 | 15000 | 3.9302 |
2.5971 | 7.71 | 15500 | 3.9282 |
2.5998 | 7.96 | 16000 | 3.9255 |
2.4553 | 8.2 | 16500 | 3.9576 |
2.4324 | 8.45 | 17000 | 3.9625 |
2.4419 | 8.7 | 17500 | 3.9650 |
2.4425 | 8.95 | 18000 | 3.9658 |
2.3711 | 9.2 | 18500 | 3.9750 |
2.3625 | 9.45 | 19000 | 3.9769 |
2.3632 | 9.7 | 19500 | 3.9773 |
2.3609 | 9.95 | 20000 | 3.9775 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.11.0+cu113
- Datasets 2.13.0
- Tokenizers 0.13.3