deit-small-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: 0.8041
• Accuracy: 0.7701

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
0.5677	1.0	164	1.4098	0.2484
0.5304	2.0	328	1.3733	0.2852
0.3987	3.0	492	0.9506	0.6659
0.319	4.0	656	0.8480	0.7419
0.2805	5.0	820	0.8182	0.7766
0.2523	6.0	984	0.8773	0.7267
0.2383	7.0	1148	0.8355	0.7473
0.2343	8.0	1312	0.8041	0.7701
0.2327	9.0	1476	0.8225	0.7527
0.2211	10.0	1640	0.8185	0.7538
0.2209	11.0	1804	0.8246	0.7570
0.2246	12.0	1968	0.8262	0.7603
0.2163	13.0	2132	0.8204	0.7516
0.2209	14.0	2296	0.8121	0.7657
0.2238	15.0	2460	0.8288	0.7636
0.2207	16.0	2624	0.8130	0.7668
0.2236	17.0	2788	0.8139	0.7625
0.2149	18.0	2952	0.8136	0.7701
0.221	19.0	3116	0.8131	0.7701
0.2147	20.0	3280	0.8131	0.7701

Framework versions

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