swin-large-patch4-window7-224-in22k-limb

This model is a fine-tuned version of microsoft/swin-large-patch4-window7-224-in22k on the c14kevincardenas/beta_caller_284 dataset. It achieves the following results on the evaluation set:

• Loss: 0.9104
• Accuracy: 0.6153

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: 32
• eval_batch_size: 32
• seed: 2014
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 30.0
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.005

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.381	1.0	215	1.3791	0.2784
1.2719	2.0	430	1.2902	0.3435
1.2584	3.0	645	1.1300	0.4423
1.1503	4.0	860	1.0864	0.5049
1.0337	5.0	1075	1.1408	0.5239
1.0062	6.0	1290	1.0178	0.5428
0.9712	7.0	1505	0.9268	0.5906
0.9287	8.0	1720	0.9192	0.5939
0.8887	9.0	1935	0.9256	0.6161
0.845	10.0	2150	0.9255	0.6096
0.8045	11.0	2365	0.9433	0.6269
0.8685	12.0	2580	0.9104	0.6153
0.806	13.0	2795	0.9112	0.6285
0.7729	14.0	3010	0.9369	0.6483
0.7582	15.0	3225	1.0296	0.6244
0.77	16.0	3440	0.9607	0.6367
0.7218	17.0	3655	1.0158	0.6326
0.7389	18.0	3870	0.9320	0.6433
0.6532	19.0	4085	0.9733	0.6639
0.6812	20.0	4300	1.0518	0.6516
0.6697	21.0	4515	1.0159	0.6409
0.6968	22.0	4730	1.0112	0.6483
0.6087	23.0	4945	1.0446	0.6639
0.6268	24.0	5160	1.0730	0.6656
0.6092	25.0	5375	1.0707	0.6680
0.5765	26.0	5590	1.0561	0.6573
0.6332	27.0	5805	1.0880	0.6614
0.6267	28.0	6020	1.1181	0.6474
0.6283	29.0	6235	1.0807	0.6507
0.5614	30.0	6450	1.0881	0.6524

Framework versions

• Transformers 4.41.0.dev0
• Pytorch 2.0.1+cu117
• Datasets 2.19.1
• Tokenizers 0.19.1