c14kevincardenas/beit-base-patch16-224_alpha0.5_temp5.0_t2

beit-base-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: 0.5472
• Accuracy: 0.7994

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
1.1991	1.0	90	1.4052	0.2826
0.9493	2.0	180	1.0380	0.5435
0.6304	3.0	270	0.6769	0.7352
0.4356	4.0	360	0.6236	0.7796
0.3103	5.0	450	0.6148	0.7668
0.2511	6.0	540	0.6114	0.7757
0.2325	7.0	630	0.5848	0.7974
0.2122	8.0	720	0.5919	0.7935
0.2001	9.0	810	0.6232	0.7806
0.1882	10.0	900	0.5793	0.7905
0.183	11.0	990	0.6269	0.7737
0.1725	12.0	1080	0.5524	0.8073
0.1628	13.0	1170	0.5641	0.7915
0.1573	14.0	1260	0.5472	0.7994
0.1579	15.0	1350	0.5699	0.7885
0.1539	16.0	1440	0.5585	0.7935
0.1547	17.0	1530	0.5541	0.7955
0.1449	18.0	1620	0.5601	0.7964
0.1463	19.0	1710	0.5563	0.7974
0.1421	20.0	1800	0.5571	0.8004

Framework versions

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