vit-base_rvl_cdip-N1K_AURC_256

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: 0.2459
• Accuracy: 0.8968
• Brier Loss: 0.1720
• Nll: 0.9246
• F1 Micro: 0.8968
• F1 Macro: 0.8967
• Ece: 0.0709
• Aurc: 0.0191

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: 256
• eval_batch_size: 256
• 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	63	0.1138	0.8922	0.1604	1.1695	0.8922	0.8926	0.0478	0.0170
No log	2.0	126	0.1565	0.8952	0.1607	1.1000	0.8952	0.8952	0.0532	0.0176
No log	3.0	189	0.1722	0.8972	0.1620	1.0250	0.8972	0.8973	0.0584	0.0175
No log	4.0	252	0.2006	0.897	0.1642	0.9921	0.897	0.8969	0.0615	0.0181
No log	5.0	315	0.2142	0.8988	0.1668	0.9670	0.8988	0.8986	0.0640	0.0183
No log	6.0	378	0.2207	0.8975	0.1688	0.9482	0.8975	0.8975	0.0674	0.0186
No log	7.0	441	0.2310	0.897	0.1700	0.9397	0.897	0.8969	0.0697	0.0188
0.008	8.0	504	0.2401	0.8968	0.1714	0.9268	0.8968	0.8966	0.0705	0.0190
0.008	9.0	567	0.2441	0.8975	0.1719	0.9262	0.8975	0.8974	0.0709	0.0191
0.008	10.0	630	0.2459	0.8968	0.1720	0.9246	0.8968	0.8967	0.0709	0.0191

Framework versions

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