layoutlmv3-finetuned-cord_100

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

• Loss: 0.2834
• Precision: 0.9175
• Recall: 0.9319
• F1: 0.9246
• Accuracy: 0.9406

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: 1e-05
• train_batch_size: 5
• eval_batch_size: 5
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• training_steps: 2500

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	4.17	250	1.0175	0.7358	0.7882	0.7611	0.8014
1.406	8.33	500	0.5646	0.8444	0.8735	0.8587	0.8671
1.406	12.5	750	0.3943	0.8950	0.9184	0.9065	0.9189
0.3467	16.67	1000	0.3379	0.9138	0.9289	0.9213	0.9291
0.3467	20.83	1250	0.2842	0.9189	0.9334	0.9261	0.9419
0.1484	25.0	1500	0.2822	0.9233	0.9371	0.9302	0.9427
0.1484	29.17	1750	0.2906	0.9168	0.9319	0.9243	0.9372
0.0825	33.33	2000	0.2922	0.9183	0.9334	0.9258	0.9410
0.0825	37.5	2250	0.2842	0.9154	0.9319	0.9236	0.9397
0.0596	41.67	2500	0.2834	0.9175	0.9319	0.9246	0.9406

Framework versions

• Transformers 4.20.1
• Pytorch 1.11.0
• Datasets 2.1.0
• Tokenizers 0.12.1