swin-tiny-patch4-window7-224_alpha0.5_temp5.0_t3

This model is a fine-tuned version of c14kevincardenas/ClimBEiT-t3 on the c14kevincardenas/beta_caller_284_person_crop_seq_withlimb dataset. It achieves the following results on the evaluation set:

• Loss: 0.6308
• Accuracy: 0.8178

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: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.8115	1.0	164	1.4105	0.2560
0.7178	2.0	328	1.1599	0.4685
0.5783	3.0	492	0.9091	0.6616
0.4814	4.0	656	0.8235	0.7137
0.4242	5.0	820	0.7623	0.7560
0.3773	6.0	984	0.7324	0.7777
0.3503	7.0	1148	0.6889	0.8080
0.3418	8.0	1312	0.6577	0.8091
0.3284	9.0	1476	0.6747	0.8015
0.3015	10.0	1640	0.6572	0.8091
0.2979	11.0	1804	0.6616	0.8156
0.3018	12.0	1968	0.6517	0.8341
0.286	13.0	2132	0.6450	0.8308
0.2976	14.0	2296	0.6335	0.8330
0.2938	15.0	2460	0.6308	0.8178
0.2894	16.0	2624	0.6356	0.8297
0.2899	17.0	2788	0.6426	0.8330
0.2837	18.0	2952	0.6312	0.8286
0.2874	19.0	3116	0.6341	0.8275
0.28	20.0	3280	0.6328	0.8351

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1