<!-- 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. -->
model_output_original_subreddit-AskScienceFiction_1
This model is a fine-tuned version of gpt2 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 3.6407
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
- gradient_accumulation_steps: 8
- total_train_batch_size: 512
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 1000
- num_epochs: 5
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 3.9231 | 0.3 | 500 | 3.8087 |
| 3.8459 | 0.6 | 1000 | 3.7766 |
| 3.8217 | 0.9 | 1500 | 3.7372 |
| 3.6939 | 1.21 | 2000 | 3.7237 |
| 3.6745 | 1.51 | 2500 | 3.7030 |
| 3.6757 | 1.81 | 3000 | 3.6811 |
| 3.5099 | 2.11 | 3500 | 3.6839 |
| 3.505 | 2.41 | 4000 | 3.6709 |
| 3.5232 | 2.71 | 4500 | 3.6515 |
| 3.3416 | 3.01 | 5000 | 3.6563 |
| 3.3725 | 3.32 | 5500 | 3.6496 |
| 3.3672 | 3.62 | 6000 | 3.6373 |
| 3.3495 | 3.92 | 6500 | 3.6280 |
| 3.2464 | 4.22 | 7000 | 3.6439 |
| 3.2467 | 4.52 | 7500 | 3.6415 |
| 3.2473 | 4.82 | 8000 | 3.6407 |
Framework versions
- Transformers 4.24.0
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2
