c14kevincardenas/swin-tiny-patch4-window7-224_alpha0.7_temp5.0_t2

swin-tiny-patch4-window7-224_alpha0.7_temp5.0

This model is a fine-tuned version of c14kevincardenas/ClimBEiTv2 on the c14kevincardenas/beta_caller_284_person_crop_seq_withlimb_2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.6171
• Accuracy: 0.7846

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: 64
• eval_batch_size: 64
• 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
1.1268	1.0	90	1.4176	0.2816
1.0188	2.0	180	1.2295	0.4328
0.7746	3.0	270	0.9900	0.5810
0.5837	4.0	360	0.7797	0.6887
0.4252	5.0	450	0.7051	0.7549
0.3312	6.0	540	0.6723	0.7628
0.2824	7.0	630	0.6632	0.7658
0.23	8.0	720	0.6645	0.7658
0.2136	9.0	810	0.6822	0.7628
0.1945	10.0	900	0.6596	0.7708
0.1714	11.0	990	0.6373	0.7747
0.1629	12.0	1080	0.6352	0.7806
0.156	13.0	1170	0.6299	0.7836
0.1515	14.0	1260	0.6171	0.7846
0.15	15.0	1350	0.6194	0.7885
0.1405	16.0	1440	0.6302	0.7816
0.1397	17.0	1530	0.6268	0.7836
0.1292	18.0	1620	0.6223	0.7796
0.1309	19.0	1710	0.6194	0.7905
0.1289	20.0	1800	0.6186	0.7935

Framework versions

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