6-classifier-finetuned-padchest

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: 0.6407
• F1: 0.7990

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: 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: 50

Training results

Training Loss	Epoch	Step	Validation Loss	F1
2.0829	1.0	18	2.0240	0.1072
1.9599	2.0	36	1.8375	0.3757
1.725	3.0	54	1.5851	0.4462
1.5014	4.0	72	1.3785	0.4928
1.3135	5.0	90	1.2678	0.5368
1.2446	6.0	108	1.1646	0.6053
1.1576	7.0	126	1.1553	0.5554
1.0868	8.0	144	1.0353	0.6231
1.0121	9.0	162	1.0081	0.6435
0.988	10.0	180	0.9306	0.6951
0.9663	11.0	198	0.9062	0.7062
0.8709	12.0	216	0.8939	0.6950
0.8891	13.0	234	0.8283	0.7371
0.843	14.0	252	0.7945	0.7482
0.8339	15.0	270	0.8384	0.7236
0.8029	16.0	288	0.8167	0.7426
0.777	17.0	306	0.7842	0.7659
0.7592	18.0	324	0.8064	0.7427
0.7052	19.0	342	0.7804	0.7553
0.7556	20.0	360	0.7332	0.7851
0.688	21.0	378	0.7643	0.7676
0.7216	22.0	396	0.7391	0.7623
0.6434	23.0	414	0.6996	0.7869
0.6673	24.0	432	0.7297	0.7775
0.6474	25.0	450	0.7006	0.7807
0.6352	26.0	468	0.7134	0.7778
0.6068	27.0	486	0.7377	0.7776
0.5942	28.0	504	0.6723	0.8089
0.5945	29.0	522	0.6686	0.7941
0.603	30.0	540	0.6667	0.7809
0.5974	31.0	558	0.6698	0.7946
0.5743	32.0	576	0.6531	0.8090
0.5663	33.0	594	0.6756	0.8013
0.5583	34.0	612	0.6535	0.8025
0.5199	35.0	630	0.6542	0.7936
0.5851	36.0	648	0.6595	0.7956
0.5105	37.0	666	0.6784	0.7886
0.4947	38.0	684	0.6625	0.8002
0.5197	39.0	702	0.6637	0.7975
0.514	40.0	720	0.6527	0.7925
0.4949	41.0	738	0.6482	0.7992
0.5047	42.0	756	0.6427	0.8036
0.5058	43.0	774	0.6437	0.8052
0.4645	44.0	792	0.6324	0.8062
0.4411	45.0	810	0.6481	0.8052
0.4602	46.0	828	0.6460	0.8037
0.4265	47.0	846	0.6505	0.8036
0.4945	48.0	864	0.6467	0.7991
0.4794	49.0	882	0.6388	0.8084
0.442	50.0	900	0.6407	0.7990

Framework versions

• Transformers 4.28.0.dev0
• Pytorch 2.0.0+cu117
• Datasets 2.18.0
• Tokenizers 0.13.3