Fer_vit_jaffe_crop_GOOGLE_1

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

• Loss: 0.2564
• Accuracy: 0.9

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: 5e-05
• train_batch_size: 32
• eval_batch_size: 32
• seed: 1
• gradient_accumulation_steps: 4
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	1	2.8373	0.1667
No log	2.0	2	2.6726	0.1667
No log	3.0	3	2.4131	0.1
No log	4.0	4	2.1618	0.1
No log	5.0	5	1.9925	0.2333
No log	6.0	6	2.0082	0.1
No log	7.0	7	2.0631	0.1667
No log	8.0	8	1.9582	0.1667
No log	9.0	9	1.9078	0.1
2.2546	10.0	10	1.8412	0.2333
2.2546	11.0	11	1.7763	0.3
2.2546	12.0	12	1.7447	0.4
2.2546	13.0	13	1.6744	0.2667
2.2546	14.0	14	1.6643	0.3333
2.2546	15.0	15	1.5688	0.4333
2.2546	16.0	16	1.5458	0.4333
2.2546	17.0	17	1.5068	0.5
2.2546	18.0	18	1.4505	0.4333
2.2546	19.0	19	1.3640	0.4
1.352	20.0	20	1.3531	0.3667
1.352	21.0	21	1.3028	0.4667
1.352	22.0	22	1.2587	0.5
1.352	23.0	23	1.3161	0.5
1.352	24.0	24	1.2402	0.4333
1.352	25.0	25	1.1801	0.4333
1.352	26.0	26	1.1508	0.5333
1.352	27.0	27	1.0463	0.6667
1.352	28.0	28	1.0176	0.5
1.352	29.0	29	1.0326	0.5333
0.6369	30.0	30	0.9021	0.6
0.6369	31.0	31	0.9485	0.5
0.6369	32.0	32	0.8393	0.6
0.6369	33.0	33	0.9536	0.5667
0.6369	34.0	34	0.8815	0.6333
0.6369	35.0	35	0.8329	0.6333
0.6369	36.0	36	0.7946	0.7333
0.6369	37.0	37	0.8582	0.6333
0.6369	38.0	38	0.7418	0.7667
0.6369	39.0	39	0.7232	0.7667
0.2532	40.0	40	0.7750	0.7333
0.2532	41.0	41	0.7209	0.7333
0.2532	42.0	42	0.6851	0.7
0.2532	43.0	43	0.6823	0.7667
0.2532	44.0	44	0.5122	0.7667
0.2532	45.0	45	0.5930	0.7667
0.2532	46.0	46	0.6531	0.7333
0.2532	47.0	47	0.5651	0.8
0.2532	48.0	48	0.5014	0.8667
0.2532	49.0	49	0.4853	0.8333
0.098	50.0	50	0.4904	0.8667
0.098	51.0	51	0.6781	0.7
0.098	52.0	52	0.6540	0.8
0.098	53.0	53	0.7150	0.7
0.098	54.0	54	0.5828	0.8
0.098	55.0	55	0.5115	0.8
0.098	56.0	56	0.4744	0.8
0.098	57.0	57	0.4548	0.8667
0.098	58.0	58	0.4936	0.8667
0.098	59.0	59	0.3534	0.8667
0.0473	60.0	60	0.6354	0.7333
0.0473	61.0	61	0.4243	0.9
0.0473	62.0	62	0.2744	0.9333
0.0473	63.0	63	0.4937	0.8333
0.0473	64.0	64	0.3869	0.9
0.0473	65.0	65	0.5379	0.8667
0.0473	66.0	66	0.4878	0.8
0.0473	67.0	67	0.6310	0.7667
0.0473	68.0	68	0.5021	0.8
0.0473	69.0	69	0.5109	0.8667
0.0218	70.0	70	0.4052	0.8667
0.0218	71.0	71	0.3340	0.9
0.0218	72.0	72	0.4823	0.8333
0.0218	73.0	73	0.2980	0.9
0.0218	74.0	74	0.3515	0.8667
0.0218	75.0	75	0.4199	0.8
0.0218	76.0	76	0.4145	0.9
0.0218	77.0	77	0.4639	0.7667
0.0218	78.0	78	0.3376	0.8667
0.0218	79.0	79	0.3546	0.8667
0.0121	80.0	80	0.3863	0.8667
0.0121	81.0	81	0.3637	0.8667
0.0121	82.0	82	0.3622	0.8667
0.0121	83.0	83	0.4142	0.8667
0.0121	84.0	84	0.4829	0.7667
0.0121	85.0	85	0.4039	0.8667
0.0121	86.0	86	0.3893	0.9
0.0121	87.0	87	0.5483	0.8333
0.0121	88.0	88	0.3928	0.8333
0.0121	89.0	89	0.3336	0.8667
0.0077	90.0	90	0.2689	0.9333
0.0077	91.0	91	0.3586	0.9333
0.0077	92.0	92	0.4284	0.9
0.0077	93.0	93	0.4150	0.8333
0.0077	94.0	94	0.2941	0.9
0.0077	95.0	95	0.2634	0.8667
0.0077	96.0	96	0.2631	0.9333
0.0077	97.0	97	0.3490	0.9333
0.0077	98.0	98	0.3602	0.9
0.0077	99.0	99	0.2326	0.9333
0.0065	100.0	100	0.2564	0.9

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0