c14kevincardenas/swin-tiny-patch4-window7-224_alpha0.7_temp3.0_t2

swin-tiny-patch4-window7-224_alpha0.7_temp3.0

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: 0.6260
• Accuracy: 0.7885

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
1.1461	1.0	90	1.4228	0.2816
1.0312	2.0	180	1.2250	0.4318
0.7723	3.0	270	1.0165	0.5652
0.5804	4.0	360	0.7593	0.7085
0.4243	5.0	450	0.7048	0.7540
0.3253	6.0	540	0.6931	0.7609
0.2822	7.0	630	0.6610	0.7638
0.229	8.0	720	0.6538	0.7787
0.212	9.0	810	0.6767	0.7688
0.1972	10.0	900	0.6680	0.7569
0.174	11.0	990	0.6353	0.7757
0.1637	12.0	1080	0.6389	0.7757
0.1558	13.0	1170	0.6260	0.7885
0.153	14.0	1260	0.6343	0.7796
0.1514	15.0	1350	0.6342	0.7767
0.1419	16.0	1440	0.6450	0.7747
0.1411	17.0	1530	0.6400	0.7826
0.1303	18.0	1620	0.6328	0.7787
0.1316	19.0	1710	0.6287	0.7866
0.1303	20.0	1800	0.6265	0.7836

Framework versions

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