hiera-finetuned-stroke-binary-ultrasound

This model is a fine-tuned version of facebook/hiera-base-224-in1k-hf on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.0098
• Accuracy: 0.9951
• F1: 0.9951
• Precision: 0.9951
• Recall: 0.9951

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: 12
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.038	0.8753	100	0.0184	0.9938	0.9938	0.9939	0.9938
0.0468	1.7440	200	0.0206	0.9926	0.9926	0.9927	0.9926
0.0445	2.6127	300	0.0225	0.9901	0.9901	0.9902	0.9901
0.0415	3.4814	400	0.0187	0.9889	0.9889	0.9889	0.9889
0.0465	4.3501	500	0.0098	0.9951	0.9951	0.9951	0.9951
0.0397	5.2188	600	0.0286	0.9901	0.9901	0.9903	0.9901
0.0257	6.0875	700	0.0188	0.9926	0.9926	0.9927	0.9926
0.0434	6.9628	800	0.0209	0.9938	0.9938	0.9939	0.9938
0.0261	7.8315	900	0.0154	0.9926	0.9926	0.9926	0.9926
0.0198	8.7002	1000	0.0094	0.9951	0.9951	0.9951	0.9951
0.0207	9.5689	1100	0.0122	0.9938	0.9938	0.9939	0.9938
0.0157	10.4376	1200	0.0101	0.9951	0.9951	0.9951	0.9951
0.0188	11.3063	1300	0.0104	0.9951	0.9951	0.9951	0.9951

Framework versions

• Transformers 4.53.1
• Pytorch 2.7.1+cu126
• Datasets 3.6.0
• Tokenizers 0.21.2