swinv2-tiny-patch4-window8-256-dmae-humeda-DAV53

This model is a fine-tuned version of microsoft/swinv2-tiny-patch4-window8-256 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.8283
• Accuracy: 0.7045

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: 3e-05
• train_batch_size: 16
• eval_batch_size: 32
• seed: 42
• gradient_accumulation_steps: 8
• total_train_batch_size: 128
• 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: 30
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	1.0	6	1.6493	0.0909
No log	2.0	12	1.5699	0.3864
No log	3.0	18	1.4384	0.4205
No log	4.0	24	1.2748	0.4091
No log	5.0	30	1.2428	0.5114
No log	6.0	36	1.0682	0.6023
No log	7.0	42	1.2919	0.5
No log	8.0	48	0.9125	0.6591
No log	9.0	54	1.0308	0.5568
No log	10.0	60	0.8505	0.6705
No log	11.0	66	0.9354	0.625
No log	12.0	72	0.8283	0.7045
No log	13.0	78	0.8508	0.6705
No log	14.0	84	0.8072	0.6477
No log	15.0	90	0.8574	0.6477
No log	16.0	96	0.8278	0.625
0.7213	17.0	102	0.8671	0.6364
0.7213	18.0	108	0.8787	0.6364
0.7213	19.0	114	0.8215	0.6818
0.7213	20.0	120	0.8018	0.6932
0.7213	21.0	126	0.8278	0.6477
0.7213	22.0	132	0.8424	0.6364
0.7213	23.0	138	0.8392	0.625
0.7213	24.0	144	0.8371	0.625
0.7213	25.0	150	0.8373	0.625

Framework versions

• Transformers 4.48.2
• Pytorch 2.5.1+cu124
• Datasets 3.2.0
• Tokenizers 0.21.0