<!-- 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. -->
gpt2_left_out_bnc_spoken
This model is a fine-tuned version of gpt2 on the generator dataset. It achieves the following results on the evaluation set:
- Loss: 3.9573
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
- 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 |
|---|---|---|---|
| 5.9941 | 0.26 | 500 | 5.0785 |
| 4.744 | 0.52 | 1000 | 4.6923 |
| 4.449 | 0.78 | 1500 | 4.4585 |
| 4.2281 | 1.04 | 2000 | 4.3108 |
| 4.0362 | 1.3 | 2500 | 4.2194 |
| 3.9581 | 1.56 | 3000 | 4.1345 |
| 3.8942 | 1.82 | 3500 | 4.0589 |
| 3.7653 | 2.08 | 4000 | 4.0146 |
| 3.6477 | 2.34 | 4500 | 3.9820 |
| 3.6314 | 2.6 | 5000 | 3.9363 |
| 3.5895 | 2.86 | 5500 | 3.8927 |
| 3.4677 | 3.12 | 6000 | 3.8892 |
| 3.3837 | 3.39 | 6500 | 3.8736 |
| 3.3922 | 3.65 | 7000 | 3.8444 |
| 3.387 | 3.91 | 7500 | 3.8169 |
| 3.2108 | 4.17 | 8000 | 3.8439 |
| 3.1722 | 4.43 | 8500 | 3.8370 |
| 3.1802 | 4.69 | 9000 | 3.8128 |
| 3.1877 | 4.95 | 9500 | 3.7892 |
| 2.9711 | 5.21 | 10000 | 3.8382 |
| 2.9515 | 5.47 | 10500 | 3.8363 |
| 2.9643 | 5.73 | 11000 | 3.8184 |
| 2.9776 | 5.99 | 11500 | 3.8051 |
| 2.7104 | 6.25 | 12000 | 3.8626 |
| 2.7359 | 6.51 | 12500 | 3.8661 |
| 2.7452 | 6.77 | 13000 | 3.8605 |
| 2.7255 | 7.03 | 13500 | 3.8748 |
| 2.5175 | 7.29 | 14000 | 3.9038 |
| 2.5252 | 7.55 | 14500 | 3.9064 |
| 2.5391 | 7.81 | 15000 | 3.9065 |
| 2.4972 | 8.07 | 15500 | 3.9270 |
| 2.3676 | 8.33 | 16000 | 3.9408 |
| 2.3852 | 8.59 | 16500 | 3.9432 |
| 2.3809 | 8.85 | 17000 | 3.9458 |
| 2.3448 | 9.11 | 17500 | 3.9530 |
| 2.2974 | 9.38 | 18000 | 3.9563 |
| 2.2979 | 9.64 | 18500 | 3.9568 |
| 2.3035 | 9.9 | 19000 | 3.9573 |
Framework versions
- Transformers 4.26.1
- Pytorch 1.11.0+cu113
- Datasets 2.13.0
- Tokenizers 0.13.3
