vit-msn-small-ultralytics_yolo_cropped_lateral_flow_ivalidation

This model is a fine-tuned version of facebook/vit-msn-small on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.4550
• Accuracy: 0.8373

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: 64
• eval_batch_size: 64
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 256
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 40

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.9032	7	0.6079	0.7263
0.4464	1.9355	15	0.4912	0.8107
0.3464	2.9677	23	0.6082	0.6820
0.2864	4.0	31	0.5636	0.7234
0.2864	4.9032	38	0.4431	0.8121
0.2617	5.9355	46	0.5066	0.7322
0.2504	6.9677	54	0.4550	0.8373
0.2319	8.0	62	0.7023	0.6686
0.2319	8.9032	69	0.6887	0.6346
0.2338	9.9355	77	0.5075	0.8107
0.2163	10.9677	85	0.6170	0.7189
0.2024	12.0	93	0.7783	0.6139
0.2027	12.9032	100	0.9525	0.5059
0.2027	13.9355	108	0.7353	0.6805
0.2086	14.9677	116	0.7734	0.6479
0.1921	16.0	124	0.9112	0.5251
0.1827	16.9032	131	0.6997	0.6997
0.1827	17.9355	139	0.7572	0.6731
0.1854	18.9677	147	0.6843	0.7041
0.172	20.0	155	0.7237	0.6997
0.1703	20.9032	162	0.7698	0.6598
0.1587	21.9355	170	0.7597	0.6420
0.1587	22.9677	178	0.8517	0.5976
0.1673	24.0	186	0.6763	0.6672
0.1474	24.9032	193	0.8353	0.6420
0.1512	25.9355	201	0.7117	0.6953
0.1512	26.9677	209	0.8383	0.6169
0.1427	28.0	217	1.0619	0.5399
0.1501	28.9032	224	0.7946	0.6760
0.1325	29.9355	232	1.0962	0.5222
0.1314	30.9677	240	0.8824	0.6183
0.1314	32.0	248	0.8409	0.6331
0.1294	32.9032	255	0.8754	0.6021
0.1204	33.9355	263	0.8036	0.6716
0.1218	34.9677	271	0.8477	0.6568
0.1218	36.0	279	0.8739	0.6331
0.1217	36.1290	280	0.8748	0.6331

Framework versions

• Transformers 4.44.2
• Pytorch 2.4.1+cu121
• Datasets 3.2.0
• Tokenizers 0.19.1