README.md

metadata

library_name: transformers
license: apache-2.0
base_model: google/vit-base-patch16-224
tags:
  - generated_from_trainer
datasets:
  - webdataset
metrics:
  - accuracy
  - f1
  - precision
  - recall
model-index:
  - name: frost-vision-v2-google_vit-base-patch16-224-v2024-11-09
    results:
      - task:
          name: Image Classification
          type: image-classification
        dataset:
          name: webdataset
          type: webdataset
          config: default
          split: train
          args: default
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.9411971830985916
          - name: F1
            type: f1
            value: 0.8485947416137806
          - name: Precision
            type: precision
            value: 0.8540145985401459
          - name: Recall
            type: recall
            value: 0.8432432432432433

frost-vision-v2-google_vit-base-patch16-224-v2024-11-09

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

• Loss: 0.1716
• Accuracy: 0.9412
• F1: 0.8486
• Precision: 0.8540
• Recall: 0.8432

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
0.2398	1.4085	100	0.2096	0.9215	0.7833	0.8502	0.7261
0.1746	2.8169	200	0.1676	0.9370	0.8362	0.8494	0.8234
0.1316	4.2254	300	0.1750	0.9282	0.8125	0.8293	0.7964
0.1305	5.6338	400	0.1671	0.9342	0.8270	0.8498	0.8054
0.1119	7.0423	500	0.1747	0.9317	0.8240	0.8300	0.8180
0.0913	8.4507	600	0.1515	0.9415	0.8505	0.8505	0.8505
0.0964	9.8592	700	0.1680	0.9377	0.8418	0.8351	0.8486
0.0659	11.2676	800	0.1891	0.9275	0.8144	0.8144	0.8144
0.0706	12.6761	900	0.1788	0.9320	0.8234	0.8364	0.8108
0.069	14.0845	1000	0.1716	0.9412	0.8486	0.8540	0.8432
0.0543	15.4930	1100	0.1847	0.9363	0.8341	0.8489	0.8198
0.0515	16.9014	1200	0.1741	0.9408	0.8470	0.8564	0.8378
0.0489	18.3099	1300	0.1793	0.9461	0.8620	0.8628	0.8613
0.0339	19.7183	1400	0.1806	0.9444	0.8569	0.8616	0.8523
0.0409	21.1268	1500	0.1784	0.9440	0.8569	0.8561	0.8577
0.0275	22.5352	1600	0.1839	0.9437	0.8548	0.8611	0.8486
0.0231	23.9437	1700	0.1865	0.9415	0.8480	0.8622	0.8342
0.0204	25.3521	1800	0.1884	0.9405	0.8482	0.8459	0.8505
0.0245	26.7606	1900	0.1935	0.9377	0.8410	0.8387	0.8432
0.0202	28.1690	2000	0.1888	0.9394	0.8456	0.8426	0.8486
0.0187	29.5775	2100	0.1914	0.9415	0.8502	0.8517	0.8486

Framework versions

• Transformers 4.44.2
• Pytorch 2.5.0+cu121
• Datasets 3.1.0
• Tokenizers 0.19.1