convnextv2-large-1k-224-finetuned-galaxy10-decals
This model is a fine-tuned version of facebook/convnextv2-large-1k-224 on the matthieulel/galaxy10_decals dataset. It achieves the following results on the evaluation set:
- Loss: 0.4479
- Accuracy: 0.8681
- Precision: 0.8670
- Recall: 0.8681
- F1: 0.8668
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 |
|---|---|---|---|---|---|---|---|
| 1.9261 | 0.99 | 62 | 1.8153 | 0.4696 | 0.5070 | 0.4696 | 0.3875 |
| 1.2684 | 2.0 | 125 | 1.1432 | 0.6793 | 0.6395 | 0.6793 | 0.6478 |
| 0.9177 | 2.99 | 187 | 0.7477 | 0.7847 | 0.7832 | 0.7847 | 0.7720 |
| 0.6937 | 4.0 | 250 | 0.5962 | 0.8168 | 0.8145 | 0.8168 | 0.8104 |
| 0.5937 | 4.99 | 312 | 0.5862 | 0.8191 | 0.8234 | 0.8191 | 0.8167 |
| 0.5921 | 6.0 | 375 | 0.5389 | 0.8365 | 0.8454 | 0.8365 | 0.8300 |
| 0.557 | 6.99 | 437 | 0.4944 | 0.8433 | 0.8478 | 0.8433 | 0.8410 |
| 0.5522 | 8.0 | 500 | 0.5022 | 0.8427 | 0.8508 | 0.8427 | 0.8416 |
| 0.5028 | 8.99 | 562 | 0.4481 | 0.8579 | 0.8610 | 0.8579 | 0.8580 |
| 0.4801 | 10.0 | 625 | 0.4360 | 0.8551 | 0.8536 | 0.8551 | 0.8527 |
| 0.4475 | 10.99 | 687 | 0.4663 | 0.8410 | 0.8423 | 0.8410 | 0.8407 |
| 0.411 | 12.0 | 750 | 0.4444 | 0.8546 | 0.8552 | 0.8546 | 0.8538 |
| 0.4173 | 12.99 | 812 | 0.4341 | 0.8613 | 0.8627 | 0.8613 | 0.8595 |
| 0.3995 | 14.0 | 875 | 0.4380 | 0.8653 | 0.8655 | 0.8653 | 0.8637 |
| 0.3657 | 14.99 | 937 | 0.4659 | 0.8625 | 0.8633 | 0.8625 | 0.8615 |
| 0.3533 | 16.0 | 1000 | 0.4600 | 0.8602 | 0.8592 | 0.8602 | 0.8585 |
| 0.3001 | 16.99 | 1062 | 0.5069 | 0.8478 | 0.8455 | 0.8478 | 0.8450 |
| 0.318 | 18.0 | 1125 | 0.4647 | 0.8574 | 0.8576 | 0.8574 | 0.8552 |
| 0.3029 | 18.99 | 1187 | 0.4479 | 0.8681 | 0.8670 | 0.8681 | 0.8668 |
| 0.2915 | 20.0 | 1250 | 0.4772 | 0.8625 | 0.8598 | 0.8625 | 0.8586 |
| 0.2742 | 20.99 | 1312 | 0.4798 | 0.8557 | 0.8538 | 0.8557 | 0.8521 |
| 0.3067 | 22.0 | 1375 | 0.4767 | 0.8602 | 0.8573 | 0.8602 | 0.8575 |
| 0.2758 | 22.99 | 1437 | 0.5099 | 0.8506 | 0.8547 | 0.8506 | 0.8516 |
| 0.2527 | 24.0 | 1500 | 0.5016 | 0.8585 | 0.8563 | 0.8585 | 0.8565 |
| 0.253 | 24.99 | 1562 | 0.4990 | 0.8625 | 0.8605 | 0.8625 | 0.8604 |
| 0.2361 | 26.0 | 1625 | 0.4903 | 0.8602 | 0.8590 | 0.8602 | 0.8591 |
| 0.2325 | 26.99 | 1687 | 0.5062 | 0.8602 | 0.8612 | 0.8602 | 0.8600 |
| 0.2448 | 28.0 | 1750 | 0.4997 | 0.8670 | 0.8648 | 0.8670 | 0.8646 |
| 0.2354 | 28.99 | 1812 | 0.4956 | 0.8608 | 0.8586 | 0.8608 | 0.8590 |
| 0.2156 | 29.76 | 1860 | 0.4970 | 0.8630 | 0.8615 | 0.8630 | 0.8617 |
Framework versions
- Transformers 4.37.2
- Pytorch 2.3.0
- Datasets 2.19.1
- Tokenizers 0.15.1