|
--- |
|
base_model: ai-forever/rugpt3medium_based_on_gpt2 |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: my_rugpt3medium_finetune |
|
results: [] |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# my_rugpt3medium_finetune |
|
|
|
This model is a fine-tuned version of [ai-forever/rugpt3medium_based_on_gpt2](https://huggingface.co/ai-forever/rugpt3medium_based_on_gpt2) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 4.3387 |
|
|
|
## 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: 1e-05 |
|
- train_batch_size: 8 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 3 |
|
- total_train_batch_size: 24 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: cosine |
|
- lr_scheduler_warmup_steps: 1000 |
|
- num_epochs: 25 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | |
|
|:-------------:|:-----:|:----:|:---------------:| |
|
| 10.916 | 0.46 | 25 | 10.6340 | |
|
| 10.3795 | 0.92 | 50 | 9.9985 | |
|
| 9.9003 | 1.38 | 75 | 9.7015 | |
|
| 9.6822 | 1.84 | 100 | 9.5795 | |
|
| 9.5804 | 2.3 | 125 | 9.5130 | |
|
| 9.5294 | 2.76 | 150 | 9.4485 | |
|
| 9.439 | 3.22 | 175 | 9.3772 | |
|
| 9.3698 | 3.68 | 200 | 9.2804 | |
|
| 9.2964 | 4.14 | 225 | 9.1746 | |
|
| 9.1945 | 4.6 | 250 | 9.0623 | |
|
| 9.0492 | 5.06 | 275 | 8.9352 | |
|
| 8.9521 | 5.52 | 300 | 8.8157 | |
|
| 8.8634 | 5.98 | 325 | 8.6838 | |
|
| 8.7197 | 6.44 | 350 | 8.5445 | |
|
| 8.6485 | 6.9 | 375 | 8.4181 | |
|
| 8.522 | 7.36 | 400 | 8.2732 | |
|
| 8.4227 | 7.82 | 425 | 8.1704 | |
|
| 8.3083 | 8.28 | 450 | 8.0290 | |
|
| 8.1897 | 8.74 | 475 | 7.8989 | |
|
| 8.0876 | 9.2 | 500 | 7.7778 | |
|
| 7.9824 | 9.66 | 525 | 7.6368 | |
|
| 7.8762 | 10.12 | 550 | 7.4974 | |
|
| 7.7408 | 10.58 | 575 | 7.3658 | |
|
| 7.6855 | 11.04 | 600 | 7.2416 | |
|
| 7.5163 | 11.5 | 625 | 7.1291 | |
|
| 7.5079 | 11.96 | 650 | 7.0295 | |
|
| 7.2873 | 12.42 | 675 | 6.8522 | |
|
| 7.2856 | 12.88 | 700 | 6.7573 | |
|
| 7.0868 | 13.34 | 725 | 6.6651 | |
|
| 7.0886 | 13.8 | 750 | 6.5239 | |
|
| 6.9283 | 14.26 | 775 | 6.3561 | |
|
| 6.8257 | 14.72 | 800 | 6.2392 | |
|
| 6.7328 | 15.18 | 825 | 6.1004 | |
|
| 6.6153 | 15.64 | 850 | 5.9846 | |
|
| 6.5824 | 16.1 | 875 | 5.8627 | |
|
| 6.3905 | 16.56 | 900 | 5.7724 | |
|
| 6.359 | 17.02 | 925 | 5.6321 | |
|
| 6.1679 | 17.48 | 950 | 5.5329 | |
|
| 6.1526 | 17.94 | 975 | 5.4058 | |
|
| 5.9604 | 18.4 | 1000 | 5.3046 | |
|
| 5.9669 | 18.87 | 1025 | 5.1939 | |
|
| 5.6807 | 19.33 | 1050 | 5.0499 | |
|
| 5.7445 | 19.79 | 1075 | 4.9479 | |
|
| 5.6578 | 20.25 | 1100 | 4.8343 | |
|
| 5.4919 | 20.71 | 1125 | 4.7547 | |
|
| 5.4427 | 21.17 | 1150 | 4.6506 | |
|
| 5.3212 | 21.63 | 1175 | 4.5628 | |
|
| 5.2953 | 22.09 | 1200 | 4.4814 | |
|
| 5.1872 | 22.55 | 1225 | 4.4373 | |
|
| 5.1285 | 23.01 | 1250 | 4.3966 | |
|
| 5.047 | 23.47 | 1275 | 4.3611 | |
|
| 5.0698 | 23.93 | 1300 | 4.3520 | |
|
| 5.1259 | 24.39 | 1325 | 4.3408 | |
|
| 4.9851 | 24.85 | 1350 | 4.3387 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.35.2 |
|
- Pytorch 2.1.0+cu121 |
|
- Datasets 2.16.0 |
|
- Tokenizers 0.15.0 |
|
|
