Modelpublic

Wiebke/results_flausch_classification_bert-base-german-cased_spelling_corrected

sourceHugging Facemitupdated 1y agoView on Hugging Face

0likes5downloads

Model Card

resultsflauschclassificationbert-base-german-casedspelling_corrected

This model is a fine-tuned version of dbmdz/bert-base-german-cased on the None dataset. It achieves the following results on the evaluation set:

—Loss: 0.2664
—Model Preparation Time: 0.0034
—Accuracy: 0.9330
—F1: 0.9327

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
—trainbatchsize: 16
—evalbatchsize: 16
—seed: 42
—optimizer: Use OptimizerNames.ADAMWTORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
—lrschedulertype: linear
—num_epochs: 3
—mixedprecisiontraining: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Model Preparation Time	Accuracy	F1
0.3023	0.2822	500	0.2438	0.0034	0.9079	0.9077
0.2634	0.5643	1000	0.2167	0.0034	0.9206	0.9206
0.2351	0.8465	1500	0.2287	0.0034	0.9252	0.9242
0.2102	1.1287	2000	0.2417	0.0034	0.9262	0.9247
0.1827	1.4108	2500	0.2265	0.0034	0.9292	0.9286
0.173	1.6930	3000	0.2377	0.0034	0.9304	0.9294
0.1756	1.9752	3500	0.2060	0.0034	0.9298	0.9294
0.1145	2.2573	4000	0.3008	0.0034	0.9326	0.9322
0.1207	2.5395	4500	0.2868	0.0034	0.9324	0.9317
0.1287	2.8217	5000	0.2664	0.0034	0.9330	0.9327

Framework versions

—Transformers 4.52.3
—Pytorch 2.6.0+cu124
—Datasets 2.14.4
—Tokenizers 0.21.1