ca3ee3bc1987accd02fef8ca3e2c8ffc

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

• Loss: 0.7054
• Data Size: 1.0
• Epoch Runtime: 2.3989
• Accuracy: 0.4375
• F1 Macro: 0.3043
• Rouge1: 0.4375
• Rouge2: 0.0
• Rougel: 0.4375
• Rougelsum: 0.4375

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	0.7444	0	0.6281	0.5156	0.4284	0.5156	0.0	0.5156	0.5156
No log	1	19	0.8946	0.0078	1.2960	0.4219	0.3361	0.4219	0.0	0.4219	0.4219
No log	2	38	0.7508	0.0156	0.8734	0.5156	0.4572	0.5156	0.0	0.5156	0.5156
No log	3	57	0.6868	0.0312	1.0611	0.5781	0.4488	0.5781	0.0	0.5781	0.5781
No log	4	76	0.6860	0.0625	1.1778	0.5625	0.36	0.5625	0.0	0.5625	0.5625
No log	5	95	0.7087	0.125	1.3469	0.4375	0.3043	0.4375	0.0	0.4375	0.4375
0.0818	6	114	0.6975	0.25	1.3548	0.5625	0.36	0.5625	0.0	0.5625	0.5625
0.0818	7	133	0.7664	0.5	1.6704	0.4375	0.3043	0.4375	0.0	0.4375	0.4375
0.5355	8.0	152	0.7054	1.0	2.3989	0.4375	0.3043	0.4375	0.0	0.4375	0.4375

Framework versions

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