layoutxlm-finetuned-kumon_case

This model is a fine-tuned version of microsoft/layoutxlm-base on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0369
• 商品コード Precision: 0.8
• 商品コード Recall: 1.0
• 商品コード F1: 0.8889
• 商品コード Number: 8
• 商品名 Precision: 0.7619
• 商品名 Recall: 0.8
• 商品名 F1: 0.7805
• 商品名 Number: 20
• 数量 Precision: 0.85
• 数量 Recall: 0.8947
• 数量 F1: 0.8718
• 数量 Number: 19
• 納期 Precision: 0.9
• 納期 Recall: 0.9
• 納期 F1: 0.9
• 納期 Number: 20
• Overall Precision: 0.8310
• Overall Recall: 0.8806
• Overall F1: 0.8551
• Overall Accuracy: 0.9917

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: 8
• eval_batch_size: 8
• seed: 0
• 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
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	商品コード Precision	商品コード Recall	商品コード F1	商品コード Number	商品名 Precision	商品名 Recall	商品名 F1	商品名 Number	数量 Precision	数量 Recall	数量 F1	数量 Number	納期 Precision	納期 Recall	納期 F1	納期 Number	Overall Precision	Overall Recall	Overall F1	Overall Accuracy
0.4475	1.0	23	0.4143	0.0	0.0	0.0	8	0.0	0.0	0.0	20	0.0	0.0	0.0	19	0.0	0.0	0.0	20	0.0	0.0	0.0	0.8947
0.1412	2.0	46	0.1207	0.0	0.0	0.0	8	0.2727	0.3	0.2857	20	1.0	0.1579	0.2727	19	0.625	0.25	0.3571	20	0.4242	0.2090	0.2800	0.9568
0.0361	3.0	69	0.0734	1.0	0.125	0.2222	8	0.6364	0.7	0.6667	20	0.8636	1.0	0.9268	19	0.9	0.9	0.9	20	0.8	0.7761	0.7879	0.9799
0.0445	4.0	92	0.0930	1.0	0.5	0.6667	8	0.5417	0.65	0.5909	20	0.7391	0.8947	0.8095	19	0.9412	0.8	0.8649	20	0.7353	0.7463	0.7407	0.9763
0.0409	5.0	115	0.0443	1.0	0.875	0.9333	8	0.7083	0.85	0.7727	20	0.7391	0.8947	0.8095	19	0.8182	0.9	0.8571	20	0.7763	0.8806	0.8252	0.9888
0.0198	6.0	138	0.0464	0.8889	1.0	0.9412	8	0.6957	0.8	0.7442	20	0.7273	0.8421	0.7805	19	0.9	0.9	0.9	20	0.7838	0.8657	0.8227	0.9888
0.0153	7.0	161	0.0460	0.8889	1.0	0.9412	8	0.75	0.75	0.75	20	0.9474	0.9474	0.9474	19	0.9474	0.9	0.9231	20	0.8806	0.8806	0.8806	0.9882
0.019	8.0	184	0.0518	0.8889	1.0	0.9412	8	0.6818	0.75	0.7143	20	0.9474	0.9474	0.9474	19	0.9	0.9	0.9	20	0.8429	0.8806	0.8613	0.9876
0.0224	9.0	207	0.0377	0.8	1.0	0.8889	8	0.7619	0.8	0.7805	20	0.85	0.8947	0.8718	19	0.9	0.9	0.9	20	0.8310	0.8806	0.8551	0.9917
0.003	10.0	230	0.0369	0.8	1.0	0.8889	8	0.7619	0.8	0.7805	20	0.85	0.8947	0.8718	19	0.9	0.9	0.9	20	0.8310	0.8806	0.8551	0.9917

Framework versions

• Transformers 4.38.2
• Pytorch 2.2.1+cu121
• Datasets 2.18.0
• Tokenizers 0.15.2