metadata

license: cc-by-nc-sa-4.0
tags:
  - generated_from_trainer
metrics:
  - precision
  - recall
  - f1
  - accuracy
model-index:
  - name: layoutlmv3-finetuned-funsd
    results: []

layoutlmv3-finetuned-funsd

This model is a fine-tuned version of microsoft/layoutlmv3-base on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.5469
Precision: 0.8633
Recall: 0.903
F1: 0.8827
Accuracy: 0.8435

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: 10
eval_batch_size: 10
seed: 42
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
training_steps: 300

Training results

Training Loss	Epoch	Step	Validation Loss	Precision	Recall	F1	Accuracy
No log	1.67	25	1.2106	0.4708	0.5275	0.4975	0.6554
No log	3.33	50	0.7854	0.7498	0.8075	0.7776	0.7687
No log	5.0	75	0.6002	0.7932	0.8455	0.8185	0.8142
No log	6.67	100	0.6523	0.7849	0.876	0.8280	0.7781
No log	8.33	125	0.5190	0.8152	0.8755	0.8443	0.8354
No log	10.0	150	0.5064	0.8315	0.888	0.8588	0.8338
No log	11.67	175	0.5342	0.8482	0.8915	0.8693	0.8344
No log	13.33	200	0.5538	0.8492	0.8925	0.8703	0.8201
No log	15.0	225	0.5336	0.8557	0.901	0.8777	0.8349
No log	16.67	250	0.5465	0.8564	0.8975	0.8765	0.8385
No log	18.33	275	0.5403	0.8580	0.9005	0.8788	0.8439
No log	20.0	300	0.5469	0.8633	0.903	0.8827	0.8435

Framework versions

Transformers 4.12.5
Pytorch 1.10.0+cu111
Datasets 2.13.2
Tokenizers 0.10.1