vit-base_rvl_cdip-N1K_aAURC_64

This model is a fine-tuned version of jordyvl/vit-base_rvl-cdip on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.4857
• Accuracy: 0.8892
• Brier Loss: 0.1843
• Nll: 0.9506
• F1 Micro: 0.8892
• F1 Macro: 0.8895
• Ece: 0.0837
• Aurc: 0.0193

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: 2e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Brier Loss	Nll	F1 Micro	F1 Macro	Ece	Aurc
No log	1.0	250	0.3824	0.888	0.1700	1.1756	0.888	0.8884	0.0548	0.0185
0.1403	2.0	500	0.3988	0.8925	0.1681	1.1230	0.8925	0.8936	0.0549	0.0199
0.1403	3.0	750	0.4099	0.8865	0.1756	1.0948	0.8865	0.8868	0.0672	0.0187
0.0442	4.0	1000	0.4297	0.8925	0.1747	1.0568	0.8925	0.8931	0.0685	0.0191
0.0442	5.0	1250	0.4467	0.8925	0.1775	1.0202	0.8925	0.8928	0.0734	0.0194
0.0119	6.0	1500	0.4612	0.8908	0.1808	0.9834	0.8907	0.8914	0.0772	0.0191
0.0119	7.0	1750	0.4762	0.8882	0.1845	0.9761	0.8882	0.8885	0.0827	0.0197
0.0062	8.0	2000	0.4763	0.892	0.1824	0.9652	0.892	0.8923	0.0789	0.0192
0.0062	9.0	2250	0.4854	0.8892	0.1844	0.9509	0.8892	0.8895	0.0834	0.0193
0.0051	10.0	2500	0.4857	0.8892	0.1843	0.9506	0.8892	0.8895	0.0837	0.0193

Framework versions

• Transformers 4.33.3
• Pytorch 2.2.0.dev20231002
• Datasets 2.7.1
• Tokenizers 0.13.3