tapt_helpfulness_base_pretraining_no_condencing
This model is a fine-tuned version of roberta-base on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 1.3379
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: 0.0001
- train_batch_size: 32
- eval_batch_size: 64
- seed: 42
- gradient_accumulation_steps: 32
- total_train_batch_size: 1024
- optimizer: Adam with betas=(0.9,0.98) and epsilon=1e-06
- lr_scheduler_type: linear
- num_epochs: 10
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 1.6371 | 1.0 | 112 | 1.4525 |
| 1.5107 | 2.0 | 225 | 1.4102 |
| 1.4871 | 2.99 | 337 | 1.3845 |
| 1.445 | 4.0 | 450 | 1.3778 |
| 1.4412 | 4.99 | 562 | 1.3483 |
| 1.4062 | 6.0 | 675 | 1.3400 |
| 1.4028 | 6.99 | 787 | 1.3190 |
| 1.3837 | 8.0 | 900 | 1.3237 |
| 1.3734 | 9.0 | 1013 | 1.3322 |
| 1.3725 | 9.95 | 1120 | 1.3289 |
Framework versions
- Transformers 4.30.2
- Pytorch 1.13.1+cu117
- Datasets 2.13.2
- Tokenizers 0.13.3