f72e6ed8d74ea54795f1d5d37656f20d

This model is a fine-tuned version of google-bert/bert-base-german-cased on the fancyzhx/dbpedia_14 dataset. It achieves the following results on the evaluation set:

• Loss: 0.1138
• Data Size: 1.0
• Epoch Runtime: 835.2868
• Accuracy: 0.9810
• F1 Macro: 0.9811
• Rouge1: 0.9810
• Rouge2: 0.0
• Rougel: 0.9811
• Rougelsum: 0.9810

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
• distributed_type: multi-GPU
• num_devices: 4
• total_train_batch_size: 32
• total_eval_batch_size: 32
• optimizer: Use adamw_torch with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: constant
• num_epochs: 50

Training results

Training Loss	Epoch	Step	Validation Loss	Data Size	Epoch Runtime	Accuracy	F1 Macro	Rouge1	Rouge2	Rougel	Rougelsum
No log	0	0	2.6608	0	30.6501	0.0646	0.0178	0.0646	0.0	0.0646	0.0646
0.3169	1	17500	0.2446	0.0078	36.5727	0.9328	0.9333	0.9328	0.0	0.9328	0.9328
0.1597	2	35000	0.1465	0.0156	42.2585	0.9614	0.9618	0.9614	0.0	0.9614	0.9613
0.0863	3	52500	0.1185	0.0312	55.2771	0.9720	0.9719	0.9720	0.0	0.9720	0.9720
0.1157	4	70000	0.0982	0.0625	79.8565	0.9767	0.9766	0.9768	0.0	0.9767	0.9767
0.0771	5	87500	0.1175	0.125	131.5180	0.9753	0.9753	0.9754	0.0	0.9753	0.9753
0.1057	6	105000	0.0854	0.25	236.4472	0.9827	0.9827	0.9827	0.0	0.9827	0.9827
0.0005	7	122500	0.0712	0.5	430.0529	0.9854	0.9853	0.9854	0.0	0.9853	0.9853
0.0575	8.0	140000	0.0875	1.0	833.3204	0.9848	0.9848	0.9848	0.0	0.9848	0.9848
0.075	9.0	157500	0.0883	1.0	835.4664	0.9836	0.9837	0.9837	0.0	0.9836	0.9836
0.0733	10.0	175000	0.0989	1.0	829.5553	0.9839	0.9839	0.9839	0.0	0.9839	0.9838
0.0891	11.0	192500	0.1138	1.0	835.2868	0.9810	0.9811	0.9810	0.0	0.9811	0.9810

Framework versions

• Transformers 4.57.0
• Pytorch 2.8.0+cu128
• Datasets 4.3.0
• Tokenizers 0.22.1