c14kevincardenas/deit-small-distilled-patch16-224_alpha0.7_temp5.0_t2

deit-small-distilled-patch16-224_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: 0.7216
• Accuracy: 0.7391

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
1.131	1.0	90	1.4281	0.2717
0.9849	2.0	180	1.2278	0.4516
0.6799	3.0	270	0.9070	0.6225
0.4254	4.0	360	0.7696	0.6957
0.2557	5.0	450	0.7668	0.7065
0.1808	6.0	540	0.8082	0.6887
0.1585	7.0	630	0.7574	0.7233
0.1465	8.0	720	0.7413	0.7312
0.1434	9.0	810	0.7646	0.7125
0.1298	10.0	900	0.7277	0.7174
0.1222	11.0	990	0.7361	0.7204
0.1211	12.0	1080	0.7389	0.7263
0.1129	13.0	1170	0.7256	0.7332
0.1098	14.0	1260	0.7216	0.7391
0.1092	15.0	1350	0.7280	0.7283
0.1073	16.0	1440	0.7258	0.7312
0.1064	17.0	1530	0.7290	0.7332
0.1004	18.0	1620	0.7240	0.7391
0.1031	19.0	1710	0.7264	0.7312
0.1002	20.0	1800	0.7258	0.7292

Framework versions

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