layoutlmv2-er-ner

This model is a fine-tuned version of renjithks/layoutlmv2-cord-ner on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.1217
• Precision: 0.7810
• Recall: 0.8085
• F1: 0.7945
• Accuracy: 0.9747

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: 5e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.0	41	0.5441	0.0	0.0	0.0	0.8851
No log	2.0	82	0.4660	0.1019	0.0732	0.0852	0.8690
No log	3.0	123	0.2506	0.4404	0.4828	0.4606	0.9240
No log	4.0	164	0.1725	0.6120	0.6076	0.6098	0.9529
No log	5.0	205	0.1387	0.7204	0.7245	0.7225	0.9671
No log	6.0	246	0.1237	0.7742	0.7747	0.7745	0.9722
No log	7.0	287	0.1231	0.7619	0.7554	0.7586	0.9697
No log	8.0	328	0.1199	0.7994	0.7719	0.7854	0.9738
No log	9.0	369	0.1197	0.7937	0.8113	0.8024	0.9741
No log	10.0	410	0.1284	0.7581	0.7597	0.7589	0.9690
No log	11.0	451	0.1172	0.7792	0.7848	0.7820	0.9738
No log	12.0	492	0.1192	0.7913	0.7970	0.7941	0.9743
0.1858	13.0	533	0.1175	0.7960	0.8006	0.7983	0.9753
0.1858	14.0	574	0.1184	0.7724	0.8034	0.7876	0.9740
0.1858	15.0	615	0.1171	0.7882	0.8142	0.8010	0.9756
0.1858	16.0	656	0.1195	0.7829	0.8070	0.7948	0.9745
0.1858	17.0	697	0.1209	0.7810	0.8006	0.7906	0.9743
0.1858	18.0	738	0.1241	0.7806	0.7963	0.7884	0.9740
0.1858	19.0	779	0.1222	0.7755	0.8027	0.7889	0.9742
0.1858	20.0	820	0.1217	0.7810	0.8085	0.7945	0.9747

Framework versions

• Transformers 4.16.2
• Pytorch 1.9.0+cu111
• Datasets 1.18.4
• Tokenizers 0.11.6