Instructions to use Pro-Coder/skin-condition-unified-v3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Pro-Coder/skin-condition-unified-v3 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-classification", model="Pro-Coder/skin-condition-unified-v3") pipe("https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/hub/parrots.png")# pip install -U transformers accelerate # Load model directly from transformers import AutoImageProcessor, AutoModelForImageClassification processor = AutoImageProcessor.from_pretrained("Pro-Coder/skin-condition-unified-v3") model = AutoModelForImageClassification.from_pretrained("Pro-Coder/skin-condition-unified-v3", device_map="auto") - Notebooks
- Google Colab
- Kaggle
skin-condition-unified-v3
This model is a fine-tuned version of facebook/convnext-base-224-22k on the None dataset. It achieves the following results on the evaluation set:
- Loss: 1.1323
- Accuracy: 0.8269
- F1 Macro: 0.7457
- Recall Melanoma Dermatoscopic Ham10000: 0.8088
- Recall Actinic Keratosis Dermatoscopic Ham10000: 0.925
- Recall Basal Cell Carcinoma Dermatoscopic Ham10: 0.9180
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: 16
- eval_batch_size: 32
- seed: 42
- gradient_accumulation_steps: 2
- total_train_batch_size: 32
- optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 1029
- num_epochs: 15
- mixed_precision_training: Native AMP
Training results
| Training Loss | Epoch | Step | Validation Loss | Accuracy | F1 Macro | Recall Melanoma Dermatoscopic Ham10000 | Recall Actinic Keratosis Dermatoscopic Ham10000 | Recall Basal Cell Carcinoma Dermatoscopic Ham10 |
|---|---|---|---|---|---|---|---|---|
| 2.1917 | 1.0 | 686 | 1.6544 | 0.6001 | 0.4260 | 0.7721 | 0.725 | 0.8197 |
| 1.8492 | 2.0 | 1372 | 1.4693 | 0.6829 | 0.5384 | 0.6029 | 0.85 | 0.8033 |
| 1.6261 | 3.0 | 2058 | 1.3676 | 0.7268 | 0.6032 | 0.3676 | 0.85 | 0.8852 |
| 1.5171 | 4.0 | 2744 | 1.3227 | 0.7387 | 0.6489 | 0.8897 | 0.875 | 0.8689 |
| 1.3606 | 5.0 | 3430 | 1.2689 | 0.7633 | 0.6737 | 0.8603 | 0.9 | 0.9016 |
| 1.2200 | 6.0 | 4116 | 1.2079 | 0.7925 | 0.7014 | 0.7353 | 0.95 | 0.9016 |
| 1.2422 | 7.0 | 4802 | 1.2236 | 0.7961 | 0.7105 | 0.8456 | 0.875 | 0.9180 |
| 1.1893 | 8.0 | 5488 | 1.1745 | 0.8142 | 0.7310 | 0.8015 | 0.9 | 0.8852 |
| 1.1331 | 9.0 | 6174 | 1.1716 | 0.8150 | 0.7293 | 0.8456 | 0.925 | 0.9180 |
| 1.1910 | 10.0 | 6860 | 1.1494 | 0.8187 | 0.7302 | 0.8309 | 0.85 | 0.9344 |
| 0.9840 | 11.0 | 7546 | 1.1422 | 0.8257 | 0.7407 | 0.8824 | 0.875 | 0.9180 |
| 1.1288 | 12.0 | 8232 | 1.1372 | 0.8244 | 0.7379 | 0.8088 | 0.9 | 0.9180 |
| 1.1447 | 13.0 | 8918 | 1.1349 | 0.8261 | 0.7446 | 0.8015 | 0.925 | 0.9180 |
| 1.1192 | 14.0 | 9604 | 1.1318 | 0.8277 | 0.7471 | 0.8235 | 0.95 | 0.9180 |
| 1.1806 | 15.0 | 10290 | 1.1312 | 0.8277 | 0.7471 | 0.8088 | 0.925 | 0.9180 |
Framework versions
- Transformers 5.23.0
- Pytorch 2.14.1+cu130
- Datasets 5.2.0
- Tokenizers 0.23.3
- Downloads last month
- 35
Model tree for Pro-Coder/skin-condition-unified-v3
Base model
facebook/convnext-base-224-22k