vit-msn-small-beta-fia-individually-enhanced_test_1

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: 0.7134
• Accuracy: 0.8028

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: 0.2
• num_epochs: 50
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.5714	1	1.8493	0.1268
No log	1.7143	3	1.7331	0.1268
No log	2.8571	5	1.4744	0.1268
No log	4.0	7	1.1333	0.2958
No log	4.5714	8	0.9675	0.6690
1.2028	5.7143	10	0.7486	0.8028
1.2028	6.8571	12	0.7134	0.8028
1.2028	8.0	14	0.7447	0.7958
1.2028	8.5714	15	0.7686	0.8028
1.2028	9.7143	17	0.7991	0.8028
1.2028	10.8571	19	0.7821	0.8028
0.7435	12.0	21	0.7589	0.8028
0.7435	12.5714	22	0.7497	0.8028
0.7435	13.7143	24	0.7412	0.8028
0.7435	14.8571	26	0.7626	0.8028
0.7435	16.0	28	0.7868	0.8028
0.7435	16.5714	29	0.7903	0.7958
0.6624	17.7143	31	0.7725	0.7958
0.6624	18.8571	33	0.7456	0.7887
0.6624	20.0	35	0.7403	0.8099
0.6624	20.5714	36	0.7444	0.8239
0.6624	21.7143	38	0.7515	0.8099
0.6099	22.8571	40	0.7579	0.8099
0.6099	24.0	42	0.7629	0.8099
0.6099	24.5714	43	0.7650	0.8169
0.6099	25.7143	45	0.7621	0.8169
0.6099	26.8571	47	0.7571	0.8099
0.6099	28.0	49	0.7551	0.8099
0.5882	28.5714	50	0.7551	0.8099

Framework versions

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