finetuned-beit-limb-seq-t4-8heads-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_4 dataset. It achieves the following results on the evaluation set:

• Loss: 0.3936
• Accuracy: 0.9186

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
0.7398	1.0	148	0.5190	0.8778
0.4307	2.0	296	0.4325	0.9006
0.5113	3.0	444	0.4381	0.9102
0.4097	4.0	592	0.4231	0.9006
0.4298	5.0	740	0.3942	0.9198
0.4204	6.0	888	0.4213	0.9042
0.4149	7.0	1036	0.4032	0.9102
0.4028	8.0	1184	0.3973	0.9162
0.4611	9.0	1332	0.3936	0.9186
0.3807	10.0	1480	0.3946	0.9186
0.3918	11.0	1628	0.3983	0.9198
0.3929	12.0	1776	0.4013	0.9150
0.368	13.0	1924	0.4185	0.9090
0.3921	14.0	2072	0.4193	0.9102
0.3706	15.0	2220	0.4134	0.9138
0.3639	16.0	2368	0.3938	0.9198
0.3335	17.0	2516	0.4057	0.9138
0.374	18.0	2664	0.4089	0.9174
0.3723	19.0	2812	0.3999	0.9174
0.3055	20.0	2960	0.4020	0.9174

Framework versions

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