gpt2_small_AR_superbpe_32768_parallel3-100_42
This model was trained from scratch on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 4.1643
- Accuracy: 0.2994
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: 128
- eval_batch_size: 128
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- lr_scheduler_warmup_steps: 1000
- num_epochs: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|---|
| 5.1946 | 1.0 | 6475 | 4.9709 | 0.2290 |
| 4.7428 | 2.0 | 12950 | 4.5972 | 0.2597 |
| 4.5332 | 3.0 | 19425 | 4.4416 | 0.2728 |
| 4.4212 | 4.0 | 25900 | 4.3481 | 0.2815 |
| 4.3362 | 5.0 | 32375 | 4.2888 | 0.2868 |
| 4.274 | 6.0 | 38850 | 4.2470 | 0.2907 |
| 4.2219 | 7.0 | 45325 | 4.2138 | 0.2945 |
| 4.1811 | 8.0 | 51800 | 4.1884 | 0.2970 |
| 4.1449 | 9.0 | 58275 | 4.1728 | 0.2987 |
| 4.1191 | 10.0 | 64750 | 4.1643 | 0.2994 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.8.0+cu128
- Datasets 4.1.1
- Tokenizers 0.19.1
