|
--- |
|
base_model: apple/mobilevitv2-1.0-imagenet1k-256 |
|
datasets: |
|
- webdataset |
|
library_name: transformers |
|
license: other |
|
metrics: |
|
- accuracy |
|
- f1 |
|
- precision |
|
- recall |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: mobilevitv2-1.0-imagenet1k-256-finetuned_v2024-10-21-frost |
|
results: |
|
- task: |
|
type: image-classification |
|
name: Image Classification |
|
dataset: |
|
name: webdataset |
|
type: webdataset |
|
config: default |
|
split: train |
|
args: default |
|
metrics: |
|
- type: accuracy |
|
value: 0.9444444444444444 |
|
name: Accuracy |
|
- type: f1 |
|
value: 0.8544819557625145 |
|
name: F1 |
|
- type: precision |
|
value: 0.8615023474178404 |
|
name: Precision |
|
- type: recall |
|
value: 0.8475750577367206 |
|
name: Recall |
|
--- |
|
|
|
<!-- 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. --> |
|
|
|
# mobilevitv2-1.0-imagenet1k-256-finetuned_v2024-10-21-frost |
|
|
|
This model is a fine-tuned version of [apple/mobilevitv2-1.0-imagenet1k-256](https://huggingface.co/apple/mobilevitv2-1.0-imagenet1k-256) on the webdataset dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.1539 |
|
- Accuracy: 0.9444 |
|
- F1: 0.8545 |
|
- Precision: 0.8615 |
|
- Recall: 0.8476 |
|
|
|
## 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: 0.0002 |
|
- train_batch_size: 16 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- 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 |
|
- mixed_precision_training: Native AMP |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 | Precision | Recall | |
|
|:-------------:|:-------:|:----:|:---------------:|:--------:|:------:|:---------:|:------:| |
|
| 0.6635 | 1.7544 | 100 | 0.6513 | 0.7604 | 0.5705 | 0.4355 | 0.8268 | |
|
| 0.4461 | 3.5088 | 200 | 0.3972 | 0.8769 | 0.7292 | 0.6322 | 0.8614 | |
|
| 0.2599 | 5.2632 | 300 | 0.2404 | 0.9227 | 0.8049 | 0.7821 | 0.8291 | |
|
| 0.2074 | 7.0175 | 400 | 0.1942 | 0.9347 | 0.8256 | 0.8488 | 0.8037 | |
|
| 0.167 | 8.7719 | 500 | 0.1772 | 0.9364 | 0.8354 | 0.8326 | 0.8383 | |
|
| 0.1661 | 10.5263 | 600 | 0.1653 | 0.9342 | 0.8259 | 0.8417 | 0.8106 | |
|
| 0.1603 | 12.2807 | 700 | 0.1649 | 0.9409 | 0.8473 | 0.8425 | 0.8522 | |
|
| 0.1523 | 14.0351 | 800 | 0.1568 | 0.9467 | 0.8592 | 0.8735 | 0.8453 | |
|
| 0.1506 | 15.7895 | 900 | 0.1548 | 0.9431 | 0.8494 | 0.8657 | 0.8337 | |
|
| 0.1485 | 17.5439 | 1000 | 0.1539 | 0.9444 | 0.8545 | 0.8615 | 0.8476 | |
|
| 0.1263 | 19.2982 | 1100 | 0.1521 | 0.944 | 0.8535 | 0.8595 | 0.8476 | |
|
| 0.1444 | 21.0526 | 1200 | 0.1552 | 0.9418 | 0.8471 | 0.8561 | 0.8383 | |
|
| 0.1133 | 22.8070 | 1300 | 0.1531 | 0.9449 | 0.8561 | 0.8601 | 0.8522 | |
|
| 0.1019 | 24.5614 | 1400 | 0.1577 | 0.9431 | 0.8491 | 0.8675 | 0.8314 | |
|
| 0.1141 | 26.3158 | 1500 | 0.1560 | 0.9413 | 0.8472 | 0.8492 | 0.8453 | |
|
| 0.1087 | 28.0702 | 1600 | 0.1573 | 0.9422 | 0.8492 | 0.8531 | 0.8453 | |
|
| 0.1015 | 29.8246 | 1700 | 0.1545 | 0.9422 | 0.8488 | 0.8548 | 0.8430 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.44.2 |
|
- Pytorch 2.4.1+cu121 |
|
- Datasets 3.0.1 |
|
- Tokenizers 0.19.1 |
|
|
