swinv2-tiny-patch4-window16-256-finetuned-galaxy10-decals
This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window16-256 on the matthieulel/galaxy10_decals dataset. It achieves the following results on the evaluation set:
- Loss: 0.4595
- Accuracy: 0.8551
- Precision: 0.8536
- Recall: 0.8551
- F1: 0.8518
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: 5e-05
- train_batch_size: 64
- eval_batch_size: 64
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 256
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 30
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | Precision | Recall | F1 |
|---|---|---|---|---|---|---|---|
| 1.723 | 0.99 | 62 | 1.4631 | 0.4803 | 0.5152 | 0.4803 | 0.4359 |
| 1.1597 | 2.0 | 125 | 0.9498 | 0.6759 | 0.6942 | 0.6759 | 0.6657 |
| 0.9305 | 2.99 | 187 | 0.6600 | 0.7728 | 0.7592 | 0.7728 | 0.7620 |
| 0.7634 | 4.0 | 250 | 0.6276 | 0.7875 | 0.7831 | 0.7875 | 0.7765 |
| 0.6924 | 4.99 | 312 | 0.5762 | 0.7943 | 0.7972 | 0.7943 | 0.7934 |
| 0.6992 | 6.0 | 375 | 0.5421 | 0.8123 | 0.8128 | 0.8123 | 0.8059 |
| 0.6731 | 6.99 | 437 | 0.5244 | 0.8129 | 0.8153 | 0.8129 | 0.8108 |
| 0.6274 | 8.0 | 500 | 0.5279 | 0.8055 | 0.8140 | 0.8055 | 0.8019 |
| 0.6096 | 8.99 | 562 | 0.4737 | 0.8354 | 0.8336 | 0.8354 | 0.8321 |
| 0.5906 | 10.0 | 625 | 0.4792 | 0.8382 | 0.8382 | 0.8382 | 0.8357 |
| 0.5839 | 10.99 | 687 | 0.5093 | 0.8224 | 0.8322 | 0.8224 | 0.8199 |
| 0.5478 | 12.0 | 750 | 0.4601 | 0.8433 | 0.8429 | 0.8433 | 0.8411 |
| 0.5678 | 12.99 | 812 | 0.5018 | 0.8269 | 0.8322 | 0.8269 | 0.8233 |
| 0.5586 | 14.0 | 875 | 0.4503 | 0.8439 | 0.8444 | 0.8439 | 0.8423 |
| 0.5267 | 14.99 | 937 | 0.4492 | 0.8444 | 0.8416 | 0.8444 | 0.8424 |
| 0.5143 | 16.0 | 1000 | 0.4543 | 0.8484 | 0.8458 | 0.8484 | 0.8442 |
| 0.4608 | 16.99 | 1062 | 0.4616 | 0.8427 | 0.8419 | 0.8427 | 0.8398 |
| 0.4914 | 18.0 | 1125 | 0.4477 | 0.8501 | 0.8501 | 0.8501 | 0.8479 |
| 0.4889 | 18.99 | 1187 | 0.4738 | 0.8337 | 0.8383 | 0.8337 | 0.8310 |
| 0.4943 | 20.0 | 1250 | 0.4758 | 0.8388 | 0.8373 | 0.8388 | 0.8352 |
| 0.4759 | 20.99 | 1312 | 0.4550 | 0.8478 | 0.8484 | 0.8478 | 0.8456 |
| 0.49 | 22.0 | 1375 | 0.4529 | 0.8512 | 0.8520 | 0.8512 | 0.8489 |
| 0.4546 | 22.99 | 1437 | 0.4567 | 0.8472 | 0.8456 | 0.8472 | 0.8447 |
| 0.4638 | 24.0 | 1500 | 0.4598 | 0.8450 | 0.8438 | 0.8450 | 0.8431 |
| 0.4591 | 24.99 | 1562 | 0.4655 | 0.8529 | 0.8539 | 0.8529 | 0.8507 |
| 0.413 | 26.0 | 1625 | 0.4512 | 0.8546 | 0.8526 | 0.8546 | 0.8514 |
| 0.4268 | 26.99 | 1687 | 0.4511 | 0.8517 | 0.8506 | 0.8517 | 0.8496 |
| 0.4497 | 28.0 | 1750 | 0.4595 | 0.8551 | 0.8536 | 0.8551 | 0.8518 |
| 0.4183 | 28.99 | 1812 | 0.4556 | 0.8540 | 0.8532 | 0.8540 | 0.8512 |
| 0.4211 | 29.76 | 1860 | 0.4567 | 0.8529 | 0.8523 | 0.8529 | 0.8503 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.3.0
- Datasets 2.19.1
- Tokenizers 0.15.1