LayoutLMv3_jordyvl_rvl_cdip_100_examples_per_class_2023-07-07_baseline

This model is a fine-tuned version of microsoft/layoutlmv3-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.9950
• Accuracy: 0.78

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: 6
• eval_batch_size: 2
• seed: 42
• gradient_accumulation_steps: 24
• total_train_batch_size: 144
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 60

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
No log	0.9	5	2.7384	0.105
No log	1.9	10	2.6717	0.14
No log	2.9	15	2.6154	0.21
No log	3.9	20	2.5111	0.245
No log	4.9	25	2.4349	0.2575
No log	5.9	30	2.3158	0.3
No log	6.9	35	2.2286	0.335
No log	7.9	40	2.1051	0.3675
No log	8.9	45	2.0340	0.41
No log	9.9	50	1.8834	0.495
No log	10.9	55	1.7616	0.5275
No log	11.9	60	1.6547	0.5575
No log	12.9	65	1.5398	0.585
No log	13.9	70	1.4345	0.615
No log	14.9	75	1.3810	0.63
No log	15.9	80	1.2689	0.685
No log	16.9	85	1.2218	0.6625
No log	17.9	90	1.1964	0.6825
No log	18.9	95	1.1421	0.6875
No log	19.9	100	1.1136	0.71
No log	20.9	105	1.0863	0.715
No log	21.9	110	1.0472	0.7075
No log	22.9	115	1.0367	0.7375
No log	23.9	120	1.0132	0.7175
No log	24.9	125	0.9760	0.7375
No log	25.9	130	0.9697	0.7275
No log	26.9	135	0.9621	0.7375
No log	27.9	140	0.9532	0.745
No log	28.9	145	0.9258	0.7475
No log	29.9	150	0.9703	0.7475
No log	30.9	155	0.9199	0.765
No log	31.9	160	0.9678	0.745
No log	32.9	165	0.9110	0.7675
No log	33.9	170	0.9723	0.755
No log	34.9	175	0.9083	0.78
No log	35.9	180	0.9427	0.76
No log	36.9	185	0.9301	0.77
No log	37.9	190	0.9318	0.765
No log	38.9	195	0.9486	0.77
No log	39.9	200	0.9676	0.755
No log	40.9	205	0.9586	0.7675
No log	41.9	210	0.9516	0.7625
No log	42.9	215	0.9796	0.7625
No log	43.9	220	0.9764	0.77
No log	44.9	225	0.9704	0.7675
No log	45.9	230	0.9842	0.775
No log	46.9	235	1.0011	0.7625
No log	47.9	240	0.9978	0.7625
No log	48.9	245	0.9873	0.775
No log	49.9	250	0.9848	0.7825
No log	50.9	255	0.9857	0.7775
No log	51.9	260	0.9975	0.775
No log	52.9	265	0.9933	0.78
No log	53.9	270	0.9840	0.78
No log	54.9	275	0.9814	0.78
No log	55.9	280	0.9860	0.775
No log	56.9	285	0.9922	0.78
No log	57.9	290	0.9949	0.78
No log	58.9	295	0.9953	0.78
No log	59.9	300	0.9950	0.78

Framework versions

• Transformers 4.26.1
• Pytorch 1.13.1.post200
• Datasets 2.9.0
• Tokenizers 0.13.2