|
|
--- |
|
|
library_name: transformers |
|
|
tags: |
|
|
- multitask-learning |
|
|
- efficientnet |
|
|
- computer-vision |
|
|
- generated_from_trainer |
|
|
model-index: |
|
|
- name: EfficientNetV1-B4-FacesMTL-EXP1 |
|
|
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. --> |
|
|
|
|
|
# EfficientNetV1-B4-FacesMTL-EXP1 |
|
|
|
|
|
This model is a fine-tuned version of EfficientNetV2-s on faces-mtl. |
|
|
It achieves the following results on the evaluation set: |
|
|
- Gender Accuracy: 0.9006 |
|
|
- Gender F1: 0.8651 |
|
|
- Age Mae: 6.7354 |
|
|
- Age Rmse: 9.1662 |
|
|
- Loss: 84.2761 |
|
|
|
|
|
## 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: 32 |
|
|
- eval_batch_size: 32 |
|
|
- seed: 42 |
|
|
- optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments |
|
|
- lr_scheduler_type: cosine |
|
|
- num_epochs: 5 |
|
|
- mixed_precision_training: Native AMP |
|
|
|
|
|
### Training results |
|
|
|
|
|
| Training Loss | Epoch | Step | Gender Accuracy | Gender F1 | Age Mae | Age Rmse | Validation Loss | |
|
|
|:-------------:|:------:|:----:|:---------------:|:---------:|:-------:|:--------:|:---------------:| |
|
|
| 221.2506 | 0.1728 | 150 | 0.8704 | 0.7993 | 10.4148 | 13.8506 | 192.2673 | |
|
|
| 163.1323 | 0.3456 | 300 | 0.8637 | 0.7814 | 9.4884 | 12.8120 | 164.4855 | |
|
|
| 188.8794 | 0.5184 | 450 | 0.8680 | 0.8299 | 9.0322 | 12.2782 | 151.1256 | |
|
|
| 147.5274 | 0.6912 | 600 | 0.8741 | 0.8056 | 8.1886 | 10.9624 | 120.4869 | |
|
|
| 121.8239 | 0.8641 | 750 | 0.8879 | 0.8379 | 7.8131 | 10.4025 | 108.5096 | |
|
|
| 110.5511 | 1.0369 | 900 | 0.8856 | 0.8386 | 7.7040 | 10.3324 | 107.0565 | |
|
|
| 116.5555 | 1.2097 | 1050 | 0.8741 | 0.8046 | 7.4726 | 9.9674 | 99.6567 | |
|
|
| 126.6334 | 1.3825 | 1200 | 0.8902 | 0.8415 | 7.5891 | 10.3014 | 106.3999 | |
|
|
| 147.6252 | 1.5553 | 1350 | 0.8911 | 0.8509 | 7.3201 | 9.8158 | 96.6354 | |
|
|
| 124.724 | 1.7281 | 1500 | 0.8951 | 0.8546 | 7.2476 | 9.6878 | 94.1328 | |
|
|
| 107.5 | 1.9009 | 1650 | 0.8897 | 0.8372 | 7.0946 | 9.4325 | 89.2502 | |
|
|
| 91.8285 | 2.0737 | 1800 | 0.8980 | 0.8612 | 7.0833 | 9.5193 | 90.9008 | |
|
|
| 94.1933 | 2.2465 | 1950 | 0.8871 | 0.8302 | 7.0344 | 9.4989 | 90.5074 | |
|
|
| 98.9504 | 2.4194 | 2100 | 0.8928 | 0.8459 | 6.9311 | 9.3160 | 87.0540 | |
|
|
| 94.4654 | 2.5922 | 2250 | 0.8977 | 0.8611 | 6.9284 | 9.3570 | 87.8282 | |
|
|
| 85.7435 | 2.7650 | 2400 | 0.8983 | 0.8634 | 6.8776 | 9.3332 | 87.3804 | |
|
|
| 125.3979 | 2.9378 | 2550 | 0.8989 | 0.8589 | 6.8158 | 9.2204 | 85.2777 | |
|
|
| 79.05 | 3.1106 | 2700 | 0.8977 | 0.8555 | 6.8892 | 9.3617 | 87.9025 | |
|
|
| 81.3652 | 3.2834 | 2850 | 0.8954 | 0.8492 | 6.7664 | 9.1391 | 83.7815 | |
|
|
| 82.3679 | 3.4562 | 3000 | 0.8989 | 0.8641 | 6.8370 | 9.2874 | 86.5219 | |
|
|
| 83.2362 | 3.6290 | 3150 | 0.8951 | 0.8620 | 6.7723 | 9.1703 | 84.3717 | |
|
|
| 80.1852 | 3.8018 | 3300 | 0.8995 | 0.8652 | 6.6909 | 9.0639 | 82.4177 | |
|
|
| 111.4015 | 3.9747 | 3450 | 0.9012 | 0.8645 | 6.7183 | 9.1129 | 83.3005 | |
|
|
| 76.6393 | 4.1475 | 3600 | 0.9018 | 0.8635 | 6.7800 | 9.1985 | 84.8666 | |
|
|
| 80.495 | 4.3203 | 3750 | 0.9023 | 0.8673 | 6.6952 | 9.0644 | 82.4208 | |
|
|
| 104.2716 | 4.4931 | 3900 | 0.9023 | 0.8655 | 6.7867 | 9.2289 | 85.4275 | |
|
|
| 77.721 | 4.6659 | 4050 | 0.9020 | 0.8674 | 6.9182 | 9.4068 | 88.7482 | |
|
|
| 70.1717 | 4.8387 | 4200 | 0.9009 | 0.8621 | 6.7354 | 9.1496 | 83.9694 | |
|
|
|
|
|
|
|
|
### Framework versions |
|
|
|
|
|
- Transformers 4.57.1 |
|
|
- Pytorch 2.9.0+cu130 |
|
|
- Datasets 4.4.1 |
|
|
- Tokenizers 0.22.1 |
|
|
|
