finetuned-AffectNet

This model is a fine-tuned version of microsoft/beit-base-patch16-224-pt22k-ft22k on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8122
• Accuracy: 0.7345

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-06
• train_batch_size: 32
• eval_batch_size: 32
• seed: 42
• 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
2.0686	1.0	163	2.0963	0.1549
1.7148	2.0	327	1.7250	0.2943
1.4591	3.0	490	1.4418	0.4204
1.3351	4.0	654	1.2648	0.5194
1.1343	5.0	817	1.0728	0.5908
1.1022	6.0	981	0.9741	0.6355
1.0476	7.0	1144	0.9203	0.6631
1.0049	8.0	1308	0.8769	0.6760
0.9561	9.0	1471	0.8438	0.6966
0.9409	10.0	1635	0.8283	0.6988
0.9419	11.0	1798	0.7867	0.7164
0.89	12.0	1962	0.7858	0.7139
0.8761	13.0	2125	0.7704	0.7147
0.8662	14.0	2289	0.7590	0.7225
0.8561	15.0	2452	0.7574	0.7199
0.8234	16.0	2616	0.7457	0.7238
0.844	17.0	2779	0.7416	0.7255
0.7908	18.0	2943	0.7485	0.7255
0.809	19.0	3106	0.7428	0.7250
0.7976	20.0	3270	0.7597	0.7203
0.7691	21.0	3433	0.7333	0.7345
0.7408	22.0	3597	0.7362	0.7246
0.7516	23.0	3760	0.7301	0.7298
0.7887	24.0	3924	0.7263	0.7332
0.7475	25.0	4087	0.7301	0.7293
0.7619	26.0	4251	0.7334	0.7298
0.7509	27.0	4414	0.7332	0.7345
0.7212	28.0	4578	0.7301	0.7367
0.7053	29.0	4741	0.7293	0.7328
0.6634	30.0	4905	0.7412	0.7298
0.677	31.0	5068	0.7221	0.7375
0.6453	32.0	5232	0.7281	0.7392
0.6961	33.0	5395	0.7280	0.7392
0.7135	34.0	5559	0.7348	0.7362
0.6871	35.0	5722	0.7334	0.7293
0.6829	36.0	5886	0.7281	0.7328
0.6742	37.0	6049	0.7332	0.7354
0.6167	38.0	6213	0.7274	0.7384
0.665	39.0	6376	0.7322	0.7311
0.6433	40.0	6540	0.7473	0.7345
0.6661	41.0	6703	0.7358	0.7341
0.6424	42.0	6867	0.7413	0.7324
0.6369	43.0	7030	0.7314	0.7414
0.611	44.0	7194	0.7325	0.7388
0.6556	45.0	7357	0.7485	0.7354
0.6524	46.0	7521	0.7434	0.7418
0.6176	47.0	7684	0.7402	0.7410
0.6142	48.0	7848	0.7480	0.7315
0.5968	49.0	8011	0.7457	0.7384
0.6132	50.0	8175	0.7514	0.7328
0.592	51.0	8338	0.7500	0.7375
0.6347	52.0	8502	0.7533	0.7345
0.5976	53.0	8665	0.7539	0.7324
0.5496	54.0	8829	0.7495	0.7388
0.5845	55.0	8992	0.7550	0.7367
0.5624	56.0	9156	0.7606	0.7362
0.5582	57.0	9319	0.7598	0.7341
0.6206	58.0	9483	0.7608	0.7345
0.5647	59.0	9646	0.7578	0.7388
0.6093	60.0	9810	0.7646	0.7358
0.5625	61.0	9973	0.7622	0.7388
0.6114	62.0	10137	0.7702	0.7324
0.5304	63.0	10300	0.7710	0.7367
0.5646	64.0	10464	0.7807	0.7298
0.5774	65.0	10627	0.7793	0.7328
0.5825	66.0	10791	0.7786	0.7375
0.5111	67.0	10954	0.7742	0.7380
0.5849	68.0	11118	0.7779	0.7349
0.5454	69.0	11281	0.7795	0.7367
0.5158	70.0	11445	0.7806	0.7345
0.5576	71.0	11608	0.7903	0.7345
0.5394	72.0	11772	0.7812	0.7380
0.5099	73.0	11935	0.7808	0.7354
0.5209	74.0	12099	0.7851	0.7319
0.5322	75.0	12262	0.7908	0.7401
0.5351	76.0	12426	0.7960	0.7306
0.5272	77.0	12589	0.7924	0.7324
0.477	78.0	12753	0.7981	0.7332
0.5186	79.0	12916	0.7942	0.7341
0.5366	80.0	13080	0.8016	0.7367
0.4809	81.0	13243	0.8014	0.7341
0.4889	82.0	13407	0.8008	0.7354
0.5287	83.0	13570	0.8010	0.7349
0.4926	84.0	13734	0.8047	0.7371
0.4989	85.0	13897	0.8046	0.7384
0.5483	86.0	14061	0.8022	0.7371
0.5157	87.0	14224	0.8055	0.7358
0.4999	88.0	14388	0.8071	0.7319
0.519	89.0	14551	0.8083	0.7362
0.4534	90.0	14715	0.8082	0.7384
0.429	91.0	14878	0.8103	0.7354
0.5073	92.0	15042	0.8116	0.7336
0.5358	93.0	15205	0.8106	0.7341
0.5049	94.0	15369	0.8111	0.7315
0.4745	95.0	15532	0.8118	0.7336
0.5052	96.0	15696	0.8104	0.7371
0.495	97.0	15859	0.8101	0.7354
0.4752	98.0	16023	0.8117	0.7349
0.4927	99.0	16186	0.8120	0.7336
0.4875	99.69	16300	0.8122	0.7345

Framework versions

• Transformers 4.35.2
• Pytorch 2.1.0+cu121
• Datasets 2.16.1
• Tokenizers 0.15.0