vit-orinoquia

This model is a fine-tuned version of google/vit-base-patch16-224-in21k on the orinoquia dataset. It achieves the following results on the evaluation set:

• Loss: 0.1021
• Accuracy: 0.9691

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
1.9968	0.0463	100	1.8490	0.4418
1.3976	0.0927	200	1.5191	0.5054
1.3472	0.1390	300	1.3085	0.6078
0.9815	0.1854	400	1.1603	0.6314
1.2055	0.2317	500	1.0710	0.6709
1.0358	0.2780	600	1.0229	0.6820
0.8788	0.3244	700	0.8523	0.7340
0.9701	0.3707	800	0.8020	0.7497
0.6715	0.4171	900	0.7216	0.7830
0.851	0.4634	1000	0.7933	0.7527
0.6638	0.5097	1100	0.6775	0.8034
0.6149	0.5561	1200	0.6193	0.8183
0.6763	0.6024	1300	0.5211	0.8462
0.6147	0.6487	1400	0.5817	0.8229
0.6746	0.6951	1500	0.4546	0.8700
0.4658	0.7414	1600	0.4779	0.8586
0.4134	0.7878	1700	0.3890	0.8854
0.4485	0.8341	1800	0.4842	0.8518
0.4662	0.8804	1900	0.3461	0.8992
0.475	0.9268	2000	0.3462	0.8968
0.2374	0.9731	2100	0.3530	0.8936
0.2639	1.0195	2200	0.3032	0.9128
0.2466	1.0658	2300	0.3104	0.9120
0.1393	1.1121	2400	0.2706	0.9244
0.1186	1.1585	2500	0.2955	0.9193
0.121	1.2048	2600	0.2699	0.9236
0.4363	1.2512	2700	0.2491	0.9323
0.3046	1.2975	2800	0.2502	0.9290
0.1064	1.3438	2900	0.2466	0.9339
0.1233	1.3902	3000	0.2184	0.9391
0.1971	1.4365	3100	0.2066	0.9426
0.0741	1.4829	3200	0.1730	0.9510
0.1206	1.5292	3300	0.1964	0.9477
0.045	1.5755	3400	0.1719	0.9515
0.0972	1.6219	3500	0.1527	0.9588
0.1798	1.6682	3600	0.1389	0.9613
0.0468	1.7146	3700	0.1267	0.9664
0.0451	1.7609	3800	0.1337	0.9645
0.0362	1.8072	3900	0.1312	0.9648
0.0546	1.8536	4000	0.1172	0.9680
0.163	1.8999	4100	0.1091	0.9694
0.0625	1.9462	4200	0.1055	0.9686
0.0725	1.9926	4300	0.1021	0.9691

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1