<!-- 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. -->
jankgpt
This model is a fine-tuned version of gpt2 on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 2.8181
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: 8
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 1000
- num_epochs: 3
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss |
---|---|---|---|
3.1021 | 0.08 | 100 | 3.0845 |
3.0967 | 0.16 | 200 | 3.0801 |
3.0836 | 0.24 | 300 | 3.0788 |
3.0587 | 0.32 | 400 | 3.0834 |
3.0541 | 0.4 | 500 | 3.0850 |
3.0305 | 0.48 | 600 | 3.0981 |
3.0016 | 0.55 | 700 | 3.1095 |
2.9926 | 0.63 | 800 | 3.1197 |
2.9714 | 0.71 | 900 | 3.1251 |
2.9671 | 0.79 | 1000 | 3.1399 |
2.9813 | 0.87 | 1100 | 3.1221 |
3.0625 | 0.95 | 1200 | 3.0943 |
3.035 | 1.03 | 1300 | 3.0655 |
2.9603 | 1.11 | 1400 | 3.0517 |
2.9562 | 1.19 | 1500 | 3.0356 |
2.927 | 1.27 | 1600 | 3.0243 |
2.936 | 1.35 | 1700 | 3.0045 |
2.9404 | 1.43 | 1800 | 2.9877 |
2.9226 | 1.51 | 1900 | 2.9740 |
2.9015 | 1.59 | 2000 | 2.9549 |
2.8955 | 1.66 | 2100 | 2.9432 |
2.8738 | 1.74 | 2200 | 2.9272 |
2.8663 | 1.82 | 2300 | 2.9138 |
2.8515 | 1.9 | 2400 | 2.8980 |
2.8433 | 1.98 | 2500 | 2.8859 |
2.7071 | 2.06 | 2600 | 2.8801 |
2.6697 | 2.14 | 2700 | 2.8721 |
2.6623 | 2.22 | 2800 | 2.8631 |
2.6561 | 2.3 | 2900 | 2.8551 |
2.6604 | 2.38 | 3000 | 2.8465 |
2.6372 | 2.46 | 3100 | 2.8402 |
2.6279 | 2.54 | 3200 | 2.8320 |
2.6209 | 2.62 | 3300 | 2.8264 |
2.6192 | 2.69 | 3400 | 2.8226 |
2.605 | 2.77 | 3500 | 2.8204 |
2.6054 | 2.85 | 3600 | 2.8187 |
2.6183 | 2.93 | 3700 | 2.8181 |
Framework versions
- Transformers 4.29.2
- Pytorch 1.13.1+rocm5.2
- Datasets 2.12.0
- Tokenizers 0.13.3