d82f7085ccddc95ffc11085c08041783

This model is a fine-tuned version of google-bert/bert-base-german-dbmdz-uncased on the nyu-mll/glue [qqp] dataset. It achieves the following results on the evaluation set:

• Loss: 0.6596
• Data Size: 1.0
• Epoch Runtime: 539.3206
• Accuracy: 0.6320
• F1 Macro: 0.3872
• Rouge1: 0.6318
• Rouge2: 0.0
• Rougel: 0.6319
• Rougelsum: 0.6317

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	0.6545	0	17.6133	0.6010	0.4870	0.6008	0.0	0.6011	0.6011
0.5995	1	11370	0.5411	0.0078	22.0320	0.7291	0.7015	0.7291	0.0	0.7291	0.7290
0.4971	2	22740	0.4390	0.0156	25.8237	0.7878	0.7710	0.7879	0.0	0.7878	0.7877
0.4163	3	34110	0.4160	0.0312	33.9354	0.8156	0.8002	0.8157	0.0	0.8155	0.8155
0.3897	4	45480	0.4592	0.0625	51.4986	0.8130	0.7941	0.8130	0.0	0.8129	0.8129
0.4537	5	56850	0.3982	0.125	83.4530	0.8151	0.8100	0.8152	0.0	0.8151	0.8150
0.5101	6	68220	0.4854	0.25	148.3178	0.8123	0.7955	0.8123	0.0	0.8123	0.8122
0.5141	7	79590	0.5086	0.5	280.1554	0.7897	0.7821	0.7899	0.0	0.7897	0.7897
0.6586	8.0	90960	0.6585	1.0	547.6003	0.6320	0.3872	0.6318	0.0	0.6319	0.6317
0.6564	9.0	102330	0.6596	1.0	539.3206	0.6320	0.3872	0.6318	0.0	0.6319	0.6317

Framework versions

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