gpt2_small_AR_superbpe_65536_parallel3_42
This model was trained from scratch on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 4.1878
- Accuracy: 0.3024
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: 64
- eval_batch_size: 64
- seed: 42
- distributed_type: multi-GPU
- num_devices: 2
- total_train_batch_size: 128
- total_eval_batch_size: 128
- 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 |
|---|
| 4.8374 | 1.0 | 22627 | 4.7311 | 0.2531 |
| 4.585 | 2.0 | 45254 | 4.5028 | 0.2717 |
| 4.4693 | 3.0 | 67881 | 4.4007 | 0.2799 |
| 4.391 | 4.0 | 90508 | 4.3393 | 0.2853 |
| 4.3361 | 5.0 | 113135 | 4.2956 | 0.2905 |
| 4.2935 | 6.0 | 135762 | 4.2619 | 0.2940 |
| 4.2567 | 7.0 | 158389 | 4.2349 | 0.2970 |
| 4.226 | 8.0 | 181016 | 4.2150 | 0.2991 |
| 4.1982 | 9.0 | 203643 | 4.1971 | 0.3011 |
| 4.1722 | 10.0 | 226270 | 4.1878 | 0.3024 |
Framework versions
- Transformers 4.41.2
- Pytorch 2.8.0+cu128
- Datasets 4.1.1
- Tokenizers 0.19.1
