metadata
license: mit
base_model: gpt2
tags:
- generated_from_trainer
model-index:
- name: polynomial_1500_5e-4
results: []
polynomial_1500_5e-4
This model is a fine-tuned version of gpt2 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 3.0244
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: 32
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 10
- total_train_batch_size: 320
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: polynomial
- lr_scheduler_warmup_steps: 250
- training_steps: 1500
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss |
|---|---|---|---|
| 9.2753 | 0.2175 | 50 | 7.6380 |
| 6.8812 | 0.4350 | 100 | 6.3560 |
| 6.0809 | 0.6525 | 150 | 5.7615 |
| 5.5903 | 0.8699 | 200 | 5.3682 |
| 5.2166 | 1.0874 | 250 | 5.0450 |
| 4.8887 | 1.3049 | 300 | 4.7443 |
| 4.6064 | 1.5224 | 350 | 4.4729 |
| 4.3711 | 1.7399 | 400 | 4.2030 |
| 4.1277 | 1.9574 | 450 | 3.9282 |
| 3.8099 | 2.1749 | 500 | 3.7320 |
| 3.6679 | 2.3923 | 550 | 3.6049 |
| 3.5513 | 2.6098 | 600 | 3.5031 |
| 3.4651 | 2.8273 | 650 | 3.4318 |
| 3.3797 | 3.0448 | 700 | 3.3644 |
| 3.2388 | 3.2623 | 750 | 3.3170 |
| 3.2017 | 3.4798 | 800 | 3.2766 |
| 3.173 | 3.6973 | 850 | 3.2426 |
| 3.143 | 3.9147 | 900 | 3.2054 |
| 3.0392 | 4.1322 | 950 | 3.1778 |
| 2.9809 | 4.3497 | 1000 | 3.1538 |
| 2.9564 | 4.5672 | 1050 | 3.1319 |
| 2.9576 | 4.7847 | 1100 | 3.1091 |
| 2.9375 | 5.0022 | 1150 | 3.0897 |
| 2.8071 | 5.2197 | 1200 | 3.0776 |
| 2.8145 | 5.4371 | 1250 | 3.0631 |
| 2.8042 | 5.6546 | 1300 | 3.0522 |
| 2.7851 | 5.8721 | 1350 | 3.0401 |
| 2.7462 | 6.0896 | 1400 | 3.0345 |
| 2.6936 | 6.3071 | 1450 | 3.0285 |
| 2.6984 | 6.5246 | 1500 | 3.0244 |
Framework versions
- Transformers 4.40.1
- Pytorch 2.3.0+cu121
- Datasets 2.19.0
- Tokenizers 0.19.1