<!-- 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_by_upvotes_positive_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: 3.9814
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: 10
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 3.7551 | 1.07 | 1000 | 3.7881 |
| 3.5181 | 2.13 | 2000 | 3.7335 |
| 3.3476 | 3.2 | 3000 | 3.7369 |
| 3.212 | 4.27 | 4000 | 3.7678 |
| 3.0517 | 5.34 | 5000 | 3.8142 |
| 2.899 | 6.4 | 6000 | 3.8666 |
| 2.7874 | 7.47 | 7000 | 3.9208 |
| 2.7247 | 8.54 | 8000 | 3.9636 |
| 2.6566 | 9.6 | 9000 | 3.9814 |
Framework versions
- Transformers 4.24.0
- Pytorch 1.13.0+cu117
- Datasets 2.7.1
- Tokenizers 0.13.2
