|
--- |
|
license: apache-2.0 |
|
base_model: microsoft/swinv2-small-patch4-window16-256 |
|
tags: |
|
- image-classification |
|
- vision |
|
- generated_from_trainer |
|
metrics: |
|
- accuracy |
|
- precision |
|
- recall |
|
- f1 |
|
model-index: |
|
- name: swinv2-small-patch4-window16-256-finetuned-galaxy10-decals |
|
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. --> |
|
|
|
# swinv2-small-patch4-window16-256-finetuned-galaxy10-decals |
|
|
|
This model is a fine-tuned version of [microsoft/swinv2-small-patch4-window16-256](https://huggingface.co/microsoft/swinv2-small-patch4-window16-256) on the matthieulel/galaxy10_decals dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.4406 |
|
- Accuracy: 0.8602 |
|
- Precision: 0.8577 |
|
- Recall: 0.8602 |
|
- F1: 0.8585 |
|
|
|
## 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 | |
|
|:-------------:|:-----:|:----:|:---------------:|:--------:|:---------:|:------:|:------:| |
|
| 1.6168 | 0.99 | 62 | 1.3397 | 0.5006 | 0.4880 | 0.5006 | 0.4599 | |
|
| 0.9396 | 2.0 | 125 | 0.7823 | 0.7463 | 0.7602 | 0.7463 | 0.7410 | |
|
| 0.782 | 2.99 | 187 | 0.5995 | 0.7948 | 0.7937 | 0.7948 | 0.7885 | |
|
| 0.6373 | 4.0 | 250 | 0.5227 | 0.8230 | 0.8192 | 0.8230 | 0.8176 | |
|
| 0.6047 | 4.99 | 312 | 0.5238 | 0.8281 | 0.8272 | 0.8281 | 0.8262 | |
|
| 0.6143 | 6.0 | 375 | 0.5091 | 0.8348 | 0.8429 | 0.8348 | 0.8298 | |
|
| 0.5805 | 6.99 | 437 | 0.4921 | 0.8264 | 0.8275 | 0.8264 | 0.8254 | |
|
| 0.5476 | 8.0 | 500 | 0.4832 | 0.8320 | 0.8409 | 0.8320 | 0.8291 | |
|
| 0.5333 | 8.99 | 562 | 0.4456 | 0.8501 | 0.8500 | 0.8501 | 0.8477 | |
|
| 0.5062 | 10.0 | 625 | 0.4493 | 0.8467 | 0.8480 | 0.8467 | 0.8457 | |
|
| 0.5001 | 10.99 | 687 | 0.4617 | 0.8450 | 0.8468 | 0.8450 | 0.8449 | |
|
| 0.4572 | 12.0 | 750 | 0.4497 | 0.8467 | 0.8450 | 0.8467 | 0.8449 | |
|
| 0.4681 | 12.99 | 812 | 0.4588 | 0.8489 | 0.8486 | 0.8489 | 0.8452 | |
|
| 0.4747 | 14.0 | 875 | 0.4281 | 0.8529 | 0.8554 | 0.8529 | 0.8508 | |
|
| 0.4283 | 14.99 | 937 | 0.4406 | 0.8602 | 0.8577 | 0.8602 | 0.8585 | |
|
| 0.4296 | 16.0 | 1000 | 0.4458 | 0.8534 | 0.8512 | 0.8534 | 0.8498 | |
|
| 0.3734 | 16.99 | 1062 | 0.4623 | 0.8416 | 0.8419 | 0.8416 | 0.8386 | |
|
| 0.3921 | 18.0 | 1125 | 0.4438 | 0.8517 | 0.8506 | 0.8517 | 0.8496 | |
|
| 0.3954 | 18.99 | 1187 | 0.4712 | 0.8467 | 0.8487 | 0.8467 | 0.8446 | |
|
| 0.3995 | 20.0 | 1250 | 0.4648 | 0.8484 | 0.8467 | 0.8484 | 0.8448 | |
|
| 0.3859 | 20.99 | 1312 | 0.4728 | 0.8495 | 0.8487 | 0.8495 | 0.8462 | |
|
| 0.4046 | 22.0 | 1375 | 0.4720 | 0.8472 | 0.8467 | 0.8472 | 0.8453 | |
|
| 0.3651 | 22.99 | 1437 | 0.4837 | 0.8416 | 0.8409 | 0.8416 | 0.8396 | |
|
| 0.3481 | 24.0 | 1500 | 0.4742 | 0.8540 | 0.8522 | 0.8540 | 0.8524 | |
|
| 0.3706 | 24.99 | 1562 | 0.4846 | 0.8478 | 0.8477 | 0.8478 | 0.8455 | |
|
| 0.3278 | 26.0 | 1625 | 0.4798 | 0.8506 | 0.8502 | 0.8506 | 0.8484 | |
|
| 0.3484 | 26.99 | 1687 | 0.4675 | 0.8529 | 0.8538 | 0.8529 | 0.8520 | |
|
| 0.3626 | 28.0 | 1750 | 0.4768 | 0.8450 | 0.8446 | 0.8450 | 0.8429 | |
|
| 0.3324 | 28.99 | 1812 | 0.4725 | 0.8484 | 0.8470 | 0.8484 | 0.8460 | |
|
| 0.3462 | 29.76 | 1860 | 0.4737 | 0.8489 | 0.8486 | 0.8489 | 0.8472 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.37.2 |
|
- Pytorch 2.3.0 |
|
- Datasets 2.19.1 |
|
- Tokenizers 0.15.1 |
|
|
