c14kevincardenas/deit-small-distilled-patch16-224_alpha0.5_temp5.0_t2

deit-small-distilled-patch16-224_alpha0.5_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: 0.7008
• Accuracy: 0.7352

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
1.2119	1.0	90	1.4199	0.2698
1.059	2.0	180	1.2413	0.4328
0.7473	3.0	270	0.8504	0.6453
0.4798	4.0	360	0.7614	0.7016
0.2936	5.0	450	0.7519	0.7134
0.2081	6.0	540	0.7559	0.7095
0.1841	7.0	630	0.7576	0.7154
0.1766	8.0	720	0.7191	0.7243
0.1721	9.0	810	0.7418	0.7223
0.1603	10.0	900	0.7023	0.7312
0.1539	11.0	990	0.7189	0.7322
0.1526	12.0	1080	0.7107	0.7441
0.144	13.0	1170	0.7010	0.7381
0.1405	14.0	1260	0.7008	0.7352
0.1401	15.0	1350	0.7008	0.7381
0.138	16.0	1440	0.7019	0.7431
0.1375	17.0	1530	0.7053	0.7391
0.1302	18.0	1620	0.7023	0.7332
0.134	19.0	1710	0.7040	0.7431
0.1305	20.0	1800	0.7038	0.7431

Framework versions

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