metadata

license: apache-2.0
base_model: google/vit-base-patch16-224-in21k
tags:
  - image-classification
  - generated_from_trainer
datasets:
  - generator
metrics:
  - accuracy
  - f1
model-index:
  - name: stool-condition-classification
    results:
      - task:
          name: Image Classification
          type: image-classification
        dataset:
          name: stool-image
          type: generator
          config: default
          split: train
          args: default
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.8171064604185623
          - name: F1
            type: f1
            value: 0.7841031149301826

stool-condition-classification

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the stool-image dataset. It achieves the following results on the evaluation set:

Loss: 0.4538
Auroc: 0.8897
Accuracy: 0.8171
Sensitivity: 0.8111
Specificty: 0.8213
F1: 0.7841

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
num_epochs: 100
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Auroc	Accuracy	Sensitivity	Specificty	F1
0.5303	0.98	100	0.4327	0.8826	0.7942	0.7191	0.8607	0.7665
0.3909	1.96	200	0.5196	0.8675	0.8047	0.8539	0.7612	0.8042
0.5328	2.94	300	0.4421	0.8864	0.8074	0.7528	0.8557	0.7859
0.4834	3.92	400	0.4721	0.8596	0.7757	0.7135	0.8308	0.7493
0.4209	4.9	500	0.4797	0.8625	0.7863	0.6798	0.8806	0.7492
0.4567	5.88	600	0.5150	0.8688	0.7942	0.6011	0.9652	0.7329

Framework versions

Transformers 4.36.1
Pytorch 2.0.1
Datasets 2.15.0
Tokenizers 0.15.0