beitv2-ena24

This model is a fine-tuned version of timm/beitv2_base_patch16_224.in1k_ft_in22k on the ena24 dataset. It achieves the following results on the evaluation set:

• Loss: 1.1010
• Accuracy: 0.6542

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: 0.0002
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 2
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.8219	0.1302	100	2.6878	0.1748
2.8794	0.2604	200	2.6517	0.1824
2.2815	0.3906	300	2.3205	0.2802
2.1651	0.5208	400	2.1680	0.3504
2.2386	0.6510	500	2.1011	0.3557
2.2215	0.7812	600	1.9812	0.3931
1.692	0.9115	700	1.8440	0.4282
1.925	1.0417	800	1.7082	0.4573
1.6596	1.1719	900	1.6649	0.4847
1.5171	1.3021	1000	1.5892	0.4863
1.1392	1.4323	1100	1.4653	0.5641
1.4759	1.5625	1200	1.3644	0.5710
1.175	1.6927	1300	1.2720	0.6099
1.0121	1.8229	1400	1.1419	0.6550
0.8563	1.9531	1500	1.1010	0.6542

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.1
• Tokenizers 0.21.1