<!-- 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. -->
model_from_berturk_1401_v4
This model is a fine-tuned version of Buseak/model_from_berturk_1401_v3 on the None dataset. It achieves the following results on the evaluation set:
- Loss: 0.0070
- Precision: 0.9983
- Recall: 0.9981
- F1: 0.9982
- Accuracy: 0.9985
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: 2e-05
- train_batch_size: 32
- eval_batch_size: 32
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- num_epochs: 15
Training results
Training Loss | Epoch | Step | Validation Loss | Precision | Recall | F1 | Accuracy |
---|---|---|---|---|---|---|---|
No log | 1.0 | 244 | 0.1500 | 0.9425 | 0.9362 | 0.9393 | 0.9575 |
No log | 2.0 | 488 | 0.1115 | 0.9570 | 0.9519 | 0.9544 | 0.9681 |
0.1776 | 3.0 | 732 | 0.0801 | 0.9700 | 0.9665 | 0.9683 | 0.9774 |
0.1776 | 4.0 | 976 | 0.0577 | 0.9792 | 0.9772 | 0.9782 | 0.9841 |
0.1024 | 5.0 | 1220 | 0.0490 | 0.9818 | 0.9809 | 0.9814 | 0.9866 |
0.1024 | 6.0 | 1464 | 0.0346 | 0.9882 | 0.9864 | 0.9873 | 0.9907 |
0.0901 | 7.0 | 1708 | 0.0254 | 0.9920 | 0.9910 | 0.9915 | 0.9935 |
0.0901 | 8.0 | 1952 | 0.0205 | 0.9935 | 0.9922 | 0.9928 | 0.9947 |
0.0617 | 9.0 | 2196 | 0.0157 | 0.9954 | 0.9947 | 0.9951 | 0.9963 |
0.0617 | 10.0 | 2440 | 0.0126 | 0.9965 | 0.9959 | 0.9962 | 0.9970 |
0.0438 | 11.0 | 2684 | 0.0110 | 0.9969 | 0.9965 | 0.9967 | 0.9975 |
0.0438 | 12.0 | 2928 | 0.0091 | 0.9976 | 0.9974 | 0.9975 | 0.9980 |
0.033 | 13.0 | 3172 | 0.0080 | 0.9979 | 0.9977 | 0.9978 | 0.9982 |
0.033 | 14.0 | 3416 | 0.0073 | 0.9982 | 0.9979 | 0.9981 | 0.9984 |
0.0264 | 15.0 | 3660 | 0.0070 | 0.9983 | 0.9981 | 0.9982 | 0.9985 |
Framework versions
- Transformers 4.25.1
- Pytorch 1.13.0+cu116
- Datasets 2.8.0
- Tokenizers 0.13.2