67078118c057b15f17ee5309efddd77b

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

• Loss: 0.5908
• Data Size: 1.0
• Epoch Runtime: 153.6105
• Accuracy: 0.7888
• F1 Macro: 0.7884
• Rouge1: 0.7884
• Rouge2: 0.0
• Rougel: 0.7888
• Rougelsum: 0.7890

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	0.7245	0	2.9227	0.5062	0.3371	0.5059	0.0	0.5062	0.5061
No log	1	3273	0.5882	0.0078	5.3513	0.6925	0.6779	0.6925	0.0	0.6926	0.6923
0.0105	2	6546	0.5352	0.0156	5.3252	0.7469	0.7380	0.7467	0.0	0.7467	0.7465
0.5382	3	9819	0.4602	0.0312	7.9835	0.7838	0.7837	0.7836	0.0	0.7836	0.7836
0.512	4	13092	0.4686	0.0625	12.6284	0.7803	0.7792	0.7803	0.0	0.7803	0.7803
0.4652	5	16365	0.4596	0.125	21.7537	0.7875	0.7838	0.7873	0.0	0.7874	0.7873
0.5062	6	19638	0.4537	0.25	41.0410	0.7903	0.7892	0.7901	0.0	0.7904	0.7903
0.4488	7	22911	0.4486	0.5	78.4194	0.8048	0.8048	0.8046	0.0	0.8050	0.8046
0.4507	8.0	26184	0.5057	1.0	153.9162	0.8017	0.8014	0.8018	0.0	0.8017	0.8018
0.4215	9.0	29457	0.5589	1.0	153.8934	0.7796	0.7774	0.7796	0.0	0.7796	0.7798
0.4558	10.0	32730	0.5007	1.0	152.8382	0.7857	0.7841	0.7857	0.0	0.7858	0.7858
0.5907	11.0	36003	0.5908	1.0	153.6105	0.7888	0.7884	0.7884	0.0	0.7888	0.7890

Framework versions

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