beit-base-patch16-224-pt22k-ft22k-finetuned-stroke-binary

BTX24/beit-base-patch16-224-pt22k-ft22k-finetuned-stroke-binary

beit-base-patch16-224-pt22k-ft22k-finetuned-stroke-binary is a image classification model from BTX24. Use it when you need a label for an image. It is set up for transformers. The card lists the license as apache-2.0.

This model is a fine-tuned version of microsoft/beit-base-patch16-224-pt22k-ft22k on an "Binary Stroke Detection" dataset. It achieves the following results on the evaluation set: - Loss: 0.2029 - Accuracy: 0.9222 - F…

Downloads · 30 days

9% of all-time downloads

All-time downloads

215

Public

Parameters

123M

18.6 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors491 MB · 100%

At a glance

Task: Image Classification
Library: transformers
License: apache-2.0
Model type: beit
Access: Public
Created: Mar 16, 2025
Updated: Apr 5, 2025
SHA: 7f9980d5

Base models

microsoft/beit-base-patch16-224-pt22k-ft22k

Datasets

BTX24/tekno21-brain-stroke-dataset-binary

Model card

From the Hugging Face model README

beit-base-patch16-224-pt22k-ft22k-finetuned-stroke-binary

This model is a fine-tuned version of microsoft/beit-base-patch16-224-pt22k-ft22k on an "Binary Stroke Detection" dataset. It achieves the following results on the evaluation set:

Loss: 0.2029
Accuracy: 0.9222
F1: 0.9214
Precision: 0.9234
Recall: 0.9222

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: 48
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
0.7256	1.2477	100	0.6913	0.5685	0.4823	0.4731	0.5685
0.6695	2.4954	200	0.6480	0.6210	0.5164	0.5987	0.6210
0.5963	3.7430	300	0.5882	0.6725	0.6118	0.6993	0.6725
0.518	4.9907	400	0.4990	0.7481	0.7167	0.7891	0.7481
0.4325	6.2477	500	0.4090	0.8073	0.7957	0.8232	0.8073
0.3848	7.4954	600	0.3703	0.8340	0.8257	0.8482	0.8340
0.3532	8.7430	700	0.3958	0.8313	0.8201	0.8564	0.8313
0.3297	9.9907	800	0.3257	0.8611	0.8558	0.8718	0.8611
0.3281	11.2477	900	0.3169	0.8666	0.8612	0.8791	0.8666
0.2938	12.4954	1000	0.2814	0.8865	0.8841	0.8900	0.8865
0.2866	13.7430	1100	0.2828	0.8869	0.8837	0.8943	0.8869
0.2884	14.9907	1200	0.2929	0.8847	0.8810	0.8936	0.8847
0.2808	16.2477	1300	0.2458	0.9014	0.8999	0.9034	0.9014
0.258	17.4954	1400	0.2351	0.9091	0.9080	0.9102	0.9091
0.2744	18.7430	1500	0.2516	0.9014	0.8994	0.9057	0.9014
0.261	19.9907	1600	0.2453	0.9068	0.9050	0.9107	0.9068
0.2519	21.2477	1700	0.2564	0.8987	0.8961	0.9051	0.8987
0.2595	22.4954	1800	0.2318	0.9095	0.9079	0.9129	0.9095
0.2548	23.7430	1900	0.2196	0.9136	0.9128	0.9142	0.9136
0.2327	24.9907	2000	0.2376	0.9068	0.9050	0.9110	0.9068
0.2563	26.2477	2100	0.2421	0.9028	0.9005	0.9083	0.9028
0.2348	27.4954	2200	0.2213	0.9109	0.9095	0.9132	0.9109
0.2427	28.7430	2300	0.2308	0.9077	0.9060	0.9116	0.9077
0.2166	29.9907	2400	0.2152	0.9141	0.9128	0.9165	0.9141
0.2345	31.2477	2500	0.2283	0.9068	0.9049	0.9114	0.9068
0.2355	32.4954	2600	0.2173	0.9118	0.9103	0.9149	0.9118
0.2291	33.7430	2700	0.2149	0.9127	0.9113	0.9155	0.9127
0.2319	34.9907	2800	0.2123	0.9141	0.9127	0.9167	0.9141
0.222	36.2477	2900	0.2053	0.9181	0.9171	0.9197	0.9181
0.2235	37.4954	3000	0.2121	0.9141	0.9127	0.9166	0.9141
0.2221	38.7430	3100	0.2013	0.9195	0.9188	0.9200	0.9195
0.2262	39.9907	3200	0.2029	0.9222	0.9214	0.9234	0.9222
0.2171	41.2477	3300	0.2075	0.9181	0.9170	0.9202	0.9181
0.2268	42.4954	3400	0.2045	0.9190	0.9180	0.9208	0.9190
0.2222	43.7430	3500	0.2050	0.9204	0.9194	0.9222	0.9204
0.2169	44.9907	3600	0.2070	0.9177	0.9165	0.9197	0.9177
0.2245	46.2477	3700	0.2064	0.9181	0.9170	0.9201	0.9181
0.2148	47.4954	3800	0.2066	0.9181	0.9170	0.9201	0.9181

Framework versions

Transformers 4.48.3
Pytorch 2.6.0+cu124
Datasets 3.4.0
Tokenizers 0.21.0

image/png

Task: Image Classification
Library: transformers
Type: beit
License: apache-2.0
Created: Mar 16, 2025
Updated: Apr 5, 2025