This model is a fine-tuned version of BTX24/hiera-finetuned-stroke-multi-to-ct on the imagefolder dataset. It achieves the following results on the evaluation set:

Loss: 0.0060
Accuracy: 0.9987
F1: 0.9987
Precision: 0.9987
Recall: 0.9987

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: 2e-05
train_batch_size: 16
eval_batch_size: 8
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 64
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: cosine_with_restarts
lr_scheduler_warmup_ratio: 0.1
num_epochs: 24
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
1.0912	1.1765	100	0.8830	0.5194	0.4534	0.5942	0.5194
0.7259	2.3529	200	0.6586	0.6942	0.6873	0.6972	0.6942
0.5083	3.5294	300	0.4742	0.7948	0.7908	0.8169	0.7948
0.3477	4.7059	400	0.2835	0.8903	0.8902	0.8918	0.8903
0.2662	5.8824	500	0.1988	0.9252	0.9251	0.9275	0.9252
0.1948	7.0588	600	0.1340	0.9510	0.9510	0.9510	0.9510
0.1468	8.2353	700	0.0793	0.9761	0.9761	0.9761	0.9761
0.1276	9.4118	800	0.0844	0.9735	0.9736	0.9741	0.9735
0.1267	10.5882	900	0.0894	0.9677	0.9678	0.9689	0.9677
0.0888	11.7647	1000	0.0393	0.9890	0.9890	0.9890	0.9890
0.0775	12.9412	1100	0.0419	0.9871	0.9871	0.9872	0.9871
0.0707	14.1176	1200	0.0347	0.9903	0.9903	0.9904	0.9903
0.0584	15.2941	1300	0.0131	0.9974	0.9974	0.9974	0.9974
0.0292	16.4706	1400	0.0175	0.9942	0.9942	0.9942	0.9942
0.0547	17.6471	1500	0.0119	0.9961	0.9961	0.9961	0.9961
0.0365	18.8235	1600	0.0076	0.9981	0.9981	0.9981	0.9981
0.0298	20.0	1700	0.0060	0.9987	0.9987	0.9987	0.9987
0.024	21.1765	1800	0.0059	0.9981	0.9981	0.9981	0.9981
0.0284	22.3529	1900	0.0059	0.9981	0.9981	0.9981	0.9981
0.0234	23.5294	2000	0.0056	0.9981	0.9981	0.9981	0.9981