finetuned-beit-limb-seq-t2-2heads-2layers-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_2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3930
• Accuracy: 0.9140

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
0.5123	1.0	180	0.4309	0.9150
0.4506	2.0	360	0.4489	0.9101
0.4381	3.0	540	0.4330	0.9002
0.4298	4.0	720	0.4092	0.9071
0.3973	5.0	900	0.4056	0.9140
0.4295	6.0	1080	0.4047	0.9140
0.4114	7.0	1260	0.4367	0.9002
0.4117	8.0	1440	0.4116	0.9170
0.4039	9.0	1620	0.4014	0.9160
0.4026	10.0	1800	0.4011	0.9130
0.3919	11.0	1980	0.3930	0.9140
0.3808	12.0	2160	0.4032	0.9121
0.371	13.0	2340	0.3959	0.9091
0.3773	14.0	2520	0.3944	0.9130
0.3334	15.0	2700	0.4254	0.9091
0.3294	16.0	2880	0.4132	0.9121
0.3222	17.0	3060	0.4043	0.9121
0.3527	18.0	3240	0.4040	0.9150
0.3269	19.0	3420	0.4080	0.9121
0.3325	20.0	3600	0.4055	0.9111

Framework versions

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