c14kevincardenas/beit-base-patch16-384_alpha0.5_temp5.0_t2

beit-base-patch16-384_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: 0.4750
• Accuracy: 0.8468

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
1.1832	1.0	90	1.3666	0.3221
0.9218	2.0	180	0.9512	0.6265
0.5975	3.0	270	0.6511	0.7668
0.4563	4.0	360	0.5916	0.7895
0.3246	5.0	450	0.5378	0.8172
0.2599	6.0	540	0.5545	0.8083
0.2359	7.0	630	0.5862	0.8063
0.2189	8.0	720	0.5557	0.8152
0.2086	9.0	810	0.5180	0.8271
0.1902	10.0	900	0.5313	0.8300
0.1912	11.0	990	0.5010	0.8261
0.1804	12.0	1080	0.5366	0.8202
0.1764	13.0	1170	0.5114	0.8340
0.1643	14.0	1260	0.5032	0.8340
0.1649	15.0	1350	0.4863	0.8419
0.1625	16.0	1440	0.4750	0.8468
0.1548	17.0	1530	0.4932	0.8498
0.1518	18.0	1620	0.4828	0.8449
0.1524	19.0	1710	0.4821	0.8439
0.1485	20.0	1800	0.4799	0.8458

Framework versions

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