|
--- |
|
license: apache-2.0 |
|
base_model: google/vit-large-patch32-384 |
|
tags: |
|
- image-classification |
|
- vision |
|
- generated_from_trainer |
|
metrics: |
|
- accuracy |
|
- precision |
|
- recall |
|
- f1 |
|
model-index: |
|
- name: vit-large-patch32-384-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. --> |
|
|
|
# vit-large-patch32-384-finetuned-galaxy10-decals |
|
|
|
This model is a fine-tuned version of [google/vit-large-patch32-384](https://huggingface.co/google/vit-large-patch32-384) on the matthieulel/galaxy10_decals dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.6766 |
|
- Accuracy: 0.8371 |
|
- Precision: 0.8374 |
|
- Recall: 0.8371 |
|
- F1: 0.8357 |
|
|
|
## 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: 0.0001 |
|
- train_batch_size: 128 |
|
- eval_batch_size: 128 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 4 |
|
- total_train_batch_size: 512 |
|
- 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 | |
|
|:-------------:|:-----:|:----:|:---------------:|:--------:|:---------:|:------:|:------:| |
|
| 1.3342 | 0.99 | 31 | 1.0491 | 0.6313 | 0.6077 | 0.6313 | 0.6052 | |
|
| 0.7979 | 1.98 | 62 | 0.6901 | 0.7672 | 0.7717 | 0.7672 | 0.7652 | |
|
| 0.7197 | 2.98 | 93 | 0.6200 | 0.7785 | 0.7716 | 0.7785 | 0.7705 | |
|
| 0.6321 | 4.0 | 125 | 0.5693 | 0.8061 | 0.8035 | 0.8061 | 0.7957 | |
|
| 0.5768 | 4.99 | 156 | 0.5501 | 0.8112 | 0.8213 | 0.8112 | 0.8134 | |
|
| 0.5173 | 5.98 | 187 | 0.5165 | 0.8213 | 0.8306 | 0.8213 | 0.8202 | |
|
| 0.4781 | 6.98 | 218 | 0.5220 | 0.8106 | 0.8161 | 0.8106 | 0.8090 | |
|
| 0.451 | 8.0 | 250 | 0.5133 | 0.8185 | 0.8227 | 0.8185 | 0.8153 | |
|
| 0.4373 | 8.99 | 281 | 0.5118 | 0.8303 | 0.8325 | 0.8303 | 0.8288 | |
|
| 0.3826 | 9.98 | 312 | 0.5280 | 0.8258 | 0.8269 | 0.8258 | 0.8243 | |
|
| 0.378 | 10.98 | 343 | 0.5477 | 0.8174 | 0.8156 | 0.8174 | 0.8142 | |
|
| 0.3509 | 12.0 | 375 | 0.5437 | 0.8281 | 0.8292 | 0.8281 | 0.8244 | |
|
| 0.3358 | 12.99 | 406 | 0.5627 | 0.8258 | 0.8268 | 0.8258 | 0.8241 | |
|
| 0.3027 | 13.98 | 437 | 0.5558 | 0.8326 | 0.8341 | 0.8326 | 0.8310 | |
|
| 0.3027 | 14.98 | 468 | 0.5703 | 0.8326 | 0.8358 | 0.8326 | 0.8295 | |
|
| 0.2786 | 16.0 | 500 | 0.5791 | 0.8281 | 0.8268 | 0.8281 | 0.8249 | |
|
| 0.2379 | 16.99 | 531 | 0.5864 | 0.8275 | 0.8264 | 0.8275 | 0.8251 | |
|
| 0.2426 | 17.98 | 562 | 0.5984 | 0.8320 | 0.8320 | 0.8320 | 0.8305 | |
|
| 0.2325 | 18.98 | 593 | 0.6217 | 0.8264 | 0.8281 | 0.8264 | 0.8252 | |
|
| 0.2208 | 20.0 | 625 | 0.6166 | 0.8258 | 0.8230 | 0.8258 | 0.8236 | |
|
| 0.2196 | 20.99 | 656 | 0.6308 | 0.8286 | 0.8280 | 0.8286 | 0.8259 | |
|
| 0.2077 | 21.98 | 687 | 0.6242 | 0.8326 | 0.8307 | 0.8326 | 0.8305 | |
|
| 0.2048 | 22.98 | 718 | 0.6801 | 0.8275 | 0.8303 | 0.8275 | 0.8263 | |
|
| 0.1886 | 24.0 | 750 | 0.6615 | 0.8264 | 0.8280 | 0.8264 | 0.8256 | |
|
| 0.2007 | 24.99 | 781 | 0.6847 | 0.8275 | 0.8280 | 0.8275 | 0.8267 | |
|
| 0.1815 | 25.98 | 812 | 0.6669 | 0.8326 | 0.8311 | 0.8326 | 0.8305 | |
|
| 0.1958 | 26.98 | 843 | 0.6766 | 0.8371 | 0.8374 | 0.8371 | 0.8357 | |
|
| 0.1806 | 28.0 | 875 | 0.6679 | 0.8360 | 0.8353 | 0.8360 | 0.8342 | |
|
| 0.1835 | 28.99 | 906 | 0.6767 | 0.8348 | 0.8334 | 0.8348 | 0.8328 | |
|
| 0.1796 | 29.76 | 930 | 0.6787 | 0.8343 | 0.8336 | 0.8343 | 0.8326 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.37.2 |
|
- Pytorch 2.3.0 |
|
- Datasets 2.19.1 |
|
- Tokenizers 0.15.1 |
|
|
