finetuned-beit-limb-seq-t4-8heads-1layers-1e-4lr

This model is a fine-tuned version of c14kevincardenas/beit-large-patch16-384-limb-person-crop on the c14kevincardenas/beta_caller_284_person_crop_seq_withlimb_4 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3927
• Accuracy: 0.9174

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: 0.0001
• train_batch_size: 32
• eval_batch_size: 32
• seed: 2014
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 250
• num_epochs: 20.0
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.05

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.0563	1.0	148	0.5863	0.8754
0.4235	2.0	296	0.4249	0.8982
0.4494	3.0	444	0.4248	0.9090
0.4183	4.0	592	0.4329	0.8934
0.4203	5.0	740	0.3979	0.9186
0.4073	6.0	888	0.3953	0.9150
0.3927	7.0	1036	0.3988	0.9210
0.3931	8.0	1184	0.4066	0.9174
0.4225	9.0	1332	0.3939	0.9222
0.3768	10.0	1480	0.4043	0.9114
0.3918	11.0	1628	0.4044	0.9102
0.398	12.0	1776	0.3927	0.9174
0.381	13.0	1924	0.3979	0.9174
0.3817	14.0	2072	0.4134	0.9138
0.3815	15.0	2220	0.4068	0.9174
0.3774	16.0	2368	0.3944	0.9234
0.3499	17.0	2516	0.3972	0.9210
0.3825	18.0	2664	0.4015	0.9198
0.3731	19.0	2812	0.3959	0.9186
0.323	20.0	2960	0.3986	0.9210

Framework versions

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