<!-- 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. -->
distilgpt2-dp
This model is a fine-tuned version of distilgpt2 on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 4.3310
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 |
---|---|---|---|
6.7548 | 0.27 | 500 | 5.6494 |
5.3558 | 0.53 | 1000 | 5.1949 |
4.9963 | 0.8 | 1500 | 4.9488 |
4.7463 | 1.07 | 2000 | 4.7983 |
4.5597 | 1.34 | 2500 | 4.6942 |
4.4758 | 1.6 | 3000 | 4.5956 |
4.3994 | 1.87 | 3500 | 4.5174 |
4.2405 | 2.14 | 4000 | 4.4796 |
4.1533 | 2.4 | 4500 | 4.4339 |
4.1326 | 2.67 | 5000 | 4.3886 |
4.1027 | 2.94 | 5500 | 4.3479 |
3.9207 | 3.21 | 6000 | 4.3582 |
3.9044 | 3.47 | 6500 | 4.3315 |
3.9109 | 3.74 | 7000 | 4.3016 |
3.8938 | 4.01 | 7500 | 4.2895 |
3.6736 | 4.27 | 8000 | 4.3069 |
3.7072 | 4.54 | 8500 | 4.2876 |
3.7151 | 4.81 | 9000 | 4.2656 |
3.6379 | 5.07 | 9500 | 4.2858 |
3.4849 | 5.34 | 10000 | 4.2893 |
3.5254 | 5.61 | 10500 | 4.2735 |
3.5283 | 5.88 | 11000 | 4.2570 |
3.3946 | 6.14 | 11500 | 4.2931 |
3.3098 | 6.41 | 12000 | 4.2970 |
3.3318 | 6.68 | 12500 | 4.2866 |
3.3499 | 6.94 | 13000 | 4.2735 |
3.1802 | 7.21 | 13500 | 4.3088 |
3.1565 | 7.48 | 14000 | 4.3098 |
3.1709 | 7.75 | 14500 | 4.3050 |
3.1565 | 8.01 | 15000 | 4.3090 |
3.021 | 8.28 | 15500 | 4.3237 |
3.0356 | 8.55 | 16000 | 4.3252 |
3.0404 | 8.81 | 16500 | 4.3249 |
3.0277 | 9.08 | 17000 | 4.3288 |
2.9656 | 9.35 | 17500 | 4.3307 |
2.977 | 9.62 | 18000 | 4.3312 |
2.9747 | 9.88 | 18500 | 4.3310 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.11.0+cu113
- Datasets 2.13.0
- Tokenizers 0.13.3