deit-tiny-distilled-patch16-224_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: 0.8429
• Accuracy: 0.7364

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
0.5593	1.0	164	1.4039	0.2560
0.5319	2.0	328	1.3999	0.2625
0.4554	3.0	492	1.0965	0.5738
0.363	4.0	656	0.9394	0.6540
0.3129	5.0	820	0.9088	0.6833
0.2708	6.0	984	0.8930	0.7007
0.2545	7.0	1148	0.8580	0.7484
0.2434	8.0	1312	0.8507	0.7343
0.2437	9.0	1476	0.8724	0.7180
0.2275	10.0	1640	0.8506	0.7278
0.2253	11.0	1804	0.8702	0.7148
0.2303	12.0	1968	0.8616	0.7364
0.2199	13.0	2132	0.8450	0.7462
0.2236	14.0	2296	0.8429	0.7364
0.2257	15.0	2460	0.8610	0.7375
0.2235	16.0	2624	0.8462	0.7397
0.2256	17.0	2788	0.8512	0.7408
0.2168	18.0	2952	0.8454	0.7397
0.2232	19.0	3116	0.8490	0.7321
0.217	20.0	3280	0.8492	0.7354

Framework versions

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