README.md

metadata

license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - imagefolder
metrics:
  - accuracy
model-index:
  - name: vc-bantai-vit-withoutAMBI-adunest-v2
    results:
      - task:
          name: Image Classification
          type: image-classification
        dataset:
          name: imagefolder
          type: imagefolder
          args: Violation-Classification---Raw-10
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.7705338809034907

vc-bantai-vit-withoutAMBI-adunest-v2

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

• Loss: 0.8271
• Accuracy: 0.7705

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: 0.0005
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 200
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.4	100	0.3811	0.8511
No log	0.81	200	0.3707	0.8609
No log	1.21	300	0.5708	0.7325
No log	1.61	400	0.3121	0.8778
0.3308	2.02	500	0.3358	0.8445
0.3308	2.42	600	0.2820	0.8768
0.3308	2.82	700	0.4825	0.7695
0.3308	3.23	800	0.3133	0.8640
0.3308	3.63	900	0.4509	0.8219
0.2028	4.03	1000	0.5426	0.7551
0.2028	4.44	1100	0.4886	0.8552
0.2028	4.84	1200	0.5649	0.7695
0.2028	5.24	1300	0.5925	0.7900
0.2028	5.65	1400	0.4203	0.8439
0.1471	6.05	1500	0.4275	0.8486
0.1471	6.45	1600	0.3683	0.8727
0.1471	6.85	1700	0.5709	0.8121
0.1471	7.26	1800	0.6209	0.7680
0.1471	7.66	1900	0.4971	0.8147
0.101	8.06	2000	0.8792	0.7567
0.101	8.47	2100	0.3288	0.8670
0.101	8.87	2200	0.3643	0.8342
0.101	9.27	2300	0.4883	0.8711
0.101	9.68	2400	0.2892	0.8943
0.0667	10.08	2500	0.5437	0.8398
0.0667	10.48	2600	0.5841	0.8450
0.0667	10.89	2700	0.8016	0.8219
0.0667	11.29	2800	0.6389	0.7772
0.0667	11.69	2900	0.3714	0.8753
0.0674	12.1	3000	0.9811	0.7130
0.0674	12.5	3100	0.6359	0.8101
0.0674	12.9	3200	0.5691	0.8285
0.0674	13.31	3300	0.6123	0.8316
0.0674	13.71	3400	0.3655	0.8978
0.0525	14.11	3500	0.4988	0.8583
0.0525	14.52	3600	0.6153	0.8450
0.0525	14.92	3700	0.4189	0.8881
0.0525	15.32	3800	0.9713	0.7967
0.0525	15.73	3900	1.1224	0.7967
0.0438	16.13	4000	0.5725	0.8578
0.0438	16.53	4100	0.4725	0.8532
0.0438	16.94	4200	0.4696	0.8640
0.0438	17.34	4300	0.4028	0.8789
0.0438	17.74	4400	0.9452	0.7746
0.0462	18.15	4500	0.4455	0.8783
0.0462	18.55	4600	0.6328	0.8311
0.0462	18.95	4700	0.6707	0.8296
0.0462	19.35	4800	0.7771	0.8429
0.0462	19.76	4900	1.2832	0.7408
0.0381	20.16	5000	0.5415	0.8737
0.0381	20.56	5100	0.8932	0.7977
0.0381	20.97	5200	0.5182	0.8691
0.0381	21.37	5300	0.5967	0.8794
0.0381	21.77	5400	0.8271	0.7705

Framework versions

• Transformers 4.20.1
• Pytorch 1.12.0+cu113
• Datasets 2.3.2
• Tokenizers 0.12.1