swin-tiny-patch4-window7-224_alpha0.7_temp3.0_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: 0.7700
• Accuracy: 0.7961

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
0.5659	1.0	164	1.4280	0.2636
0.5111	2.0	328	1.3040	0.3720
0.4093	3.0	492	0.9968	0.6356
0.3598	4.0	656	0.9535	0.6443
0.3284	5.0	820	0.8426	0.7256
0.2933	6.0	984	0.8269	0.7657
0.2734	7.0	1148	0.8815	0.6952
0.2679	8.0	1312	0.8079	0.7679
0.2574	9.0	1476	0.7823	0.7863
0.2403	10.0	1640	0.7833	0.7907
0.2376	11.0	1804	0.7851	0.7852
0.2399	12.0	1968	0.7966	0.7939
0.231	13.0	2132	0.7956	0.7766
0.2351	14.0	2296	0.7793	0.7918
0.2344	15.0	2460	0.7700	0.7961
0.232	16.0	2624	0.7845	0.7907
0.2318	17.0	2788	0.7930	0.7918
0.2241	18.0	2952	0.7814	0.7885
0.2291	19.0	3116	0.7901	0.7820
0.2227	20.0	3280	0.7854	0.7885

Framework versions

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