finetuned-beit-limb-seq-t4-2heads-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: 0.3901
• 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
1.0527	1.0	148	0.6327	0.8347
0.4279	2.0	296	0.4130	0.9102
0.4649	3.0	444	0.4122	0.9174
0.4404	4.0	592	0.4099	0.9078
0.4154	5.0	740	0.4043	0.9126
0.4252	6.0	888	0.4019	0.9114
0.405	7.0	1036	0.4010	0.9150
0.3991	8.0	1184	0.4009	0.9174
0.4304	9.0	1332	0.3901	0.9186
0.3902	10.0	1480	0.3972	0.9138
0.376	11.0	1628	0.4009	0.9174
0.4028	12.0	1776	0.3955	0.9186
0.3803	13.0	1924	0.3990	0.9186
0.4008	14.0	2072	0.4057	0.9186
0.3646	15.0	2220	0.4001	0.9174
0.3707	16.0	2368	0.3910	0.9198
0.3488	17.0	2516	0.3948	0.9198
0.3842	18.0	2664	0.3994	0.9174
0.3867	19.0	2812	0.3937	0.9174
0.3154	20.0	2960	0.3975	0.9198

Framework versions

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