finetuned-beit-limb-seq-t2-4heads-1layers-5e-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_2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3965
• Accuracy: 0.9180

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: 0.0005
• 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: 15.0
• mixed_precision_training: Native AMP
• label_smoothing_factor: 0.05

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
0.5406	1.0	180	0.4804	0.8883
0.4852	2.0	360	0.5457	0.8794
0.4664	3.0	540	0.4203	0.9051
0.4929	4.0	720	0.4349	0.9002
0.4334	5.0	900	0.4816	0.8765
0.4709	6.0	1080	0.4574	0.8933
0.4525	7.0	1260	0.4652	0.8883
0.4601	8.0	1440	0.4387	0.9002
0.4361	9.0	1620	0.4138	0.9071
0.4297	10.0	1800	0.4089	0.9140
0.4	11.0	1980	0.4000	0.9200
0.4035	12.0	2160	0.4260	0.9071
0.3875	13.0	2340	0.4088	0.9101
0.4117	14.0	2520	0.3965	0.9180
0.3518	15.0	2700	0.3987	0.9160

Framework versions

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