vit-msn-small-beta-fia-manually-enhanced-HSV_test_3

This model is a fine-tuned version of Melo1512/vit-msn-small-beta-fia-manually-enhanced-HSV_test_2 on the imagefolder dataset. It achieves the following results on the evaluation set:

• Loss: 0.5013
• Accuracy: 0.8803

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: 1e-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: 0.15
• num_epochs: 50
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.5714	1	0.5123	0.8873
No log	1.7143	3	0.5219	0.8873
No log	2.8571	5	0.5431	0.8732
No log	4.0	7	0.5444	0.8732
No log	4.5714	8	0.5336	0.8803
0.4252	5.7143	10	0.5235	0.8873
0.4252	6.8571	12	0.5269	0.8803
0.4252	8.0	14	0.5106	0.8873
0.4252	8.5714	15	0.5048	0.8873
0.4252	9.7143	17	0.5013	0.8803
0.4252	10.8571	19	0.5105	0.8803
0.4413	12.0	21	0.5256	0.8803
0.4413	12.5714	22	0.5303	0.8732
0.4413	13.7143	24	0.5218	0.8662
0.4413	14.8571	26	0.5188	0.8592
0.4413	16.0	28	0.5202	0.8592
0.4413	16.5714	29	0.5252	0.8592
0.437	17.7143	31	0.5385	0.8592
0.437	18.8571	33	0.5456	0.8592
0.437	20.0	35	0.5409	0.8732
0.437	20.5714	36	0.5375	0.8662
0.437	21.7143	38	0.5356	0.8662
0.4343	22.8571	40	0.5328	0.8803
0.4343	24.0	42	0.5318	0.8803
0.4343	24.5714	43	0.5330	0.8803
0.4343	25.7143	45	0.5334	0.8803
0.4343	26.8571	47	0.5332	0.8732
0.4343	28.0	49	0.5341	0.8732
0.4271	28.5714	50	0.5343	0.8732

Framework versions

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