README.md

metadata

license: apache-2.0
tags:
  - generated_from_trainer
datasets:
  - imagefolder
metrics:
  - accuracy
model-index:
  - name: vit-base-beans-demo-v5
    results:
      - task:
          name: Image Classification
          type: image-classification
        dataset:
          name: imagefolder
          type: imagefolder
          config: default
          split: validation
          args: default
        metrics:
          - name: Accuracy
            type: accuracy
            value: 0.46791907514450864

vit-base-beans-demo-v5

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

• Loss: 2.6708
• Accuracy: 0.4679

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: 4
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
5.9636	0.06	100	5.7983	0.1
5.8053	0.11	200	5.8683	0.1110
5.9476	0.17	300	5.9242	0.1006
5.6866	0.23	400	5.6640	0.1110
5.5886	0.29	500	5.6032	0.1153
5.4108	0.34	600	5.5314	0.1179
5.4427	0.4	700	5.4592	0.1188
5.1333	0.46	800	5.3569	0.1272
5.2427	0.52	900	5.2451	0.1318
5.2185	0.57	1000	5.1948	0.1355
4.777	0.63	1100	5.1379	0.1361
5.2378	0.69	1200	5.1043	0.1347
5.2246	0.74	1300	5.0783	0.1419
4.9846	0.8	1400	5.0425	0.1390
5.2708	0.86	1500	5.0202	0.1387
4.9169	0.92	1600	4.9382	0.1526
4.8091	0.97	1700	4.8691	0.1497
4.8795	1.03	1800	4.8124	0.1546
4.6634	1.09	1900	4.7816	0.1601
4.4967	1.15	2000	4.7105	0.1618
4.8389	1.2	2100	4.7104	0.1671
4.5872	1.26	2200	4.6636	0.1607
4.7063	1.32	2300	4.6506	0.1584
4.5526	1.38	2400	4.5932	0.1743
4.4984	1.43	2500	4.5266	0.1792
4.2266	1.49	2600	4.4860	0.1850
4.5827	1.55	2700	4.4237	0.1844
3.9383	1.6	2800	4.3919	0.1887
4.5361	1.66	2900	4.3408	0.1971
4.5067	1.72	3000	4.2708	0.1965
4.3133	1.78	3100	4.2283	0.1997
4.4104	1.83	3200	4.1830	0.2061
3.965	1.89	3300	4.1360	0.2133
4.3425	1.95	3400	4.0754	0.2237
3.9526	2.01	3500	4.0885	0.2188
3.9037	2.06	3600	3.9629	0.2396
3.6883	2.12	3700	4.0130	0.2289
3.8445	2.18	3800	3.9220	0.2540
3.6093	2.23	3900	3.9453	0.2353
3.7109	2.29	4000	3.8822	0.2402
3.588	2.35	4100	3.7765	0.2679
3.4878	2.41	4200	3.7138	0.2821
3.8276	2.46	4300	3.7137	0.2694
3.7288	2.52	4400	3.6505	0.2821
3.4948	2.58	4500	3.6280	0.2835
3.3436	2.64	4600	3.5212	0.3145
3.3389	2.69	4700	3.5006	0.3208
3.4803	2.75	4800	3.4130	0.3361
3.3953	2.81	4900	3.3506	0.3370
3.3648	2.87	5000	3.3132	0.3462
3.1838	2.92	5100	3.2632	0.3543
3.1927	2.98	5200	3.2335	0.3613
2.8337	3.04	5300	3.1633	0.3760
2.6126	3.09	5400	3.1287	0.3803
2.7718	3.15	5500	3.0715	0.3876
2.7694	3.21	5600	3.0283	0.4040
2.7131	3.27	5700	2.9859	0.4040
2.6204	3.32	5800	2.9461	0.4078
2.4889	3.38	5900	2.9413	0.4081
2.5283	3.44	6000	2.9001	0.4147
2.6986	3.5	6100	2.8428	0.4335
2.8514	3.55	6200	2.8352	0.4399
2.2355	3.61	6300	2.7825	0.4462
2.4485	3.67	6400	2.7580	0.4535
2.3359	3.72	6500	2.7330	0.4549
2.5904	3.78	6600	2.7096	0.4613
2.5366	3.84	6700	2.6906	0.4642
2.3954	3.9	6800	2.6797	0.4691
2.3722	3.95	6900	2.6708	0.4679

Framework versions

• Transformers 4.28.0
• Pytorch 2.2.1+cu121
• Datasets 2.19.1
• Tokenizers 0.13.3