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

• Loss: 0.3944
• Accuracy: 0.9239

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
0.6046	1.0	180	0.4599	0.9032
0.4623	2.0	360	0.4688	0.8982
0.4205	3.0	540	0.4379	0.9012
0.4164	4.0	720	0.4090	0.9130
0.397	5.0	900	0.4028	0.9150
0.4353	6.0	1080	0.3969	0.9140
0.4123	7.0	1260	0.4185	0.9061
0.4208	8.0	1440	0.4194	0.9071
0.4056	9.0	1620	0.3944	0.9239
0.3865	10.0	1800	0.4046	0.9140
0.3754	11.0	1980	0.4040	0.9111
0.3717	12.0	2160	0.4046	0.9111
0.3738	13.0	2340	0.4006	0.9091
0.4028	14.0	2520	0.3960	0.9130
0.3343	15.0	2700	0.4097	0.9150
0.3479	16.0	2880	0.3992	0.9140
0.3202	17.0	3060	0.4015	0.9170
0.356	18.0	3240	0.3959	0.9140
0.3412	19.0	3420	0.3967	0.9180
0.3419	20.0	3600	0.3980	0.9140

Framework versions

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