limbxy_pose_4heads_2layers_16embeddim

This model is a fine-tuned version of c14kevincardenas/beit-large-patch16-384-limb on the c14kevincardenas/beta_caller_284_limbxy_pose dataset. It achieves the following results on the evaluation set:

• Loss: 0.1391
• Rmse: 0.3730

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: 5e-05
• train_batch_size: 64
• eval_batch_size: 64
• 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

Training results

Training Loss	Epoch	Step	Validation Loss	Rmse
0.2116	1.0	89	0.1722	0.4150
0.154	2.0	178	0.1639	0.4049
0.1667	3.0	267	0.1488	0.3857
0.1487	4.0	356	0.1529	0.3910
0.153	5.0	445	0.1420	0.3769
0.1516	6.0	534	0.1413	0.3759
0.1492	7.0	623	0.1423	0.3772
0.1472	8.0	712	0.1402	0.3745
0.1507	9.0	801	0.1431	0.3783
0.1494	10.0	890	0.1399	0.3740
0.1483	11.0	979	0.1415	0.3762
0.1447	12.0	1068	0.1417	0.3764
0.1562	13.0	1157	0.1451	0.3809
0.1444	14.0	1246	0.1421	0.3770
0.1474	15.0	1335	0.1399	0.3740
0.1414	16.0	1424	0.1431	0.3783
0.1411	17.0	1513	0.1393	0.3733
0.1394	18.0	1602	0.1393	0.3732
0.1427	19.0	1691	0.1393	0.3732
0.139	20.0	1780	0.1391	0.3730

Framework versions

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