|
|
--- |
|
|
tags: |
|
|
- autogluon |
|
|
- multimodal |
|
|
- image-classification |
|
|
- resnet18 |
|
|
--- |
|
|
license: mit |
|
|
language: |
|
|
- en |
|
|
pipeline_tag: image-classification |
|
|
tags: |
|
|
- images |
|
|
datasets: |
|
|
- aedupuga/cards-image-dataset |
|
|
metrics: |
|
|
- accuracy |
|
|
- f1 |
|
|
library_name: autogluon |
|
|
|
|
|
Training Details: |
|
|
-The model was trained using AutoGluon's MultiModalPredictor with the following configuration: |
|
|
|
|
|
-Problem Type: Classification |
|
|
-Evaluation Metric: Accuracy |
|
|
-Presets: medium_quality |
|
|
-Hyperparameters: |
|
|
-model.names: ["timm_image"] |
|
|
-model.timm_image.checkpoint_name: "resnet18" |
|
|
-The training data used was the 'augmented' split of the dataset, with a 80/20 train/test split for tuning. |
|
|
|
|
|
Evaluation: |
|
|
-The model was evaluated on the 'original' split of the dataset. |
|
|
|
|
|
-Accuracy: 1.0000 |
|
|
-Weighted F1: 1.0000 |
|
|
-Note: These results are based on the evaluation performed in the provided Colab notebook. |
