<!-- 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_sorted_reversed_subreddit-wallstreetbets_1
This model is a fine-tuned version of gpt2 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 3.5417
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: 2
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 3.9986 | 0.12 | 500 | 3.8892 |
| 3.8984 | 0.25 | 1000 | 3.8108 |
| 3.8438 | 0.37 | 1500 | 3.7569 |
| 3.8001 | 0.5 | 2000 | 3.7196 |
| 3.7576 | 0.62 | 2500 | 3.6918 |
| 3.7516 | 0.75 | 3000 | 3.6667 |
| 3.7337 | 0.87 | 3500 | 3.6451 |
| 3.6981 | 1.0 | 4000 | 3.6239 |
| 3.5662 | 1.12 | 4500 | 3.6126 |
| 3.5496 | 1.25 | 5000 | 3.5951 |
| 3.5219 | 1.37 | 5500 | 3.5783 |
| 3.5422 | 1.5 | 6000 | 3.5644 |
| 3.5354 | 1.62 | 6500 | 3.5525 |
| 3.4843 | 1.75 | 7000 | 3.5462 |
| 3.52 | 1.87 | 7500 | 3.5424 |
| 3.5146 | 1.99 | 8000 | 3.5417 |
Framework versions
- Transformers 4.24.0
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2
