<!-- 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-concat
This model is a fine-tuned version of distilgpt2 on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 4.3325
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 |
|---|---|---|---|
| 6.7514 | 0.29 | 500 | 5.6224 |
| 5.3454 | 0.58 | 1000 | 5.1814 |
| 4.9931 | 0.87 | 1500 | 4.9290 |
| 4.7222 | 1.16 | 2000 | 4.7811 |
| 4.5672 | 1.45 | 2500 | 4.6657 |
| 4.4669 | 1.74 | 3000 | 4.5721 |
| 4.3738 | 2.02 | 3500 | 4.4939 |
| 4.175 | 2.31 | 4000 | 4.4613 |
| 4.1659 | 2.6 | 4500 | 4.4128 |
| 4.1369 | 2.89 | 5000 | 4.3666 |
| 3.9858 | 3.18 | 5500 | 4.3656 |
| 3.9337 | 3.47 | 6000 | 4.3419 |
| 3.9348 | 3.76 | 6500 | 4.3095 |
| 3.8826 | 4.05 | 7000 | 4.3066 |
| 3.7106 | 4.34 | 7500 | 4.3104 |
| 3.7404 | 4.63 | 8000 | 4.2893 |
| 3.7459 | 4.92 | 8500 | 4.2648 |
| 3.5695 | 5.21 | 9000 | 4.2984 |
| 3.536 | 5.49 | 9500 | 4.2887 |
| 3.5604 | 5.78 | 10000 | 4.2711 |
| 3.5007 | 6.07 | 10500 | 4.2900 |
| 3.3477 | 6.36 | 11000 | 4.3013 |
| 3.3629 | 6.65 | 11500 | 4.2906 |
| 3.3771 | 6.94 | 12000 | 4.2814 |
| 3.211 | 7.23 | 12500 | 4.3131 |
| 3.1938 | 7.52 | 13000 | 4.3124 |
| 3.21 | 7.81 | 13500 | 4.3093 |
| 3.159 | 8.1 | 14000 | 4.3204 |
| 3.0726 | 8.39 | 14500 | 4.3257 |
| 3.0762 | 8.68 | 15000 | 4.3269 |
| 3.0834 | 8.96 | 15500 | 4.3257 |
| 3.0173 | 9.25 | 16000 | 4.3311 |
| 3.0116 | 9.54 | 16500 | 4.3325 |
| 3.0155 | 9.83 | 17000 | 4.3325 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.11.0+cu113
- Datasets 2.13.0
- Tokenizers 0.13.3
