592ef1b503c686d1eccd1b3154967e2e

This model is a fine-tuned version of distilbert/distilbert-base-german-cased on the dim/tldr_news dataset. It achieves the following results on the evaluation set:

• Loss: 0.9531
• Data Size: 1.0
• Epoch Runtime: 5.7829
• Accuracy: 0.7479
• F1 Macro: 0.7743
• Rouge1: 0.7486
• Rouge2: 0.0
• Rougel: 0.7482
• Rougelsum: 0.7479

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	1.6487	0	0.9596	0.0185	0.0073	0.0185	0.0	0.0185	0.0185
No log	1	178	1.5231	0.0078	2.2349	0.3196	0.1680	0.3203	0.0	0.3200	0.3196
No log	2	356	1.4243	0.0156	1.1919	0.4545	0.2462	0.4553	0.0	0.4545	0.4538
No log	3	534	1.2987	0.0312	1.3966	0.3913	0.2289	0.3913	0.0	0.3913	0.3906
No log	4	712	1.1949	0.0625	1.5513	0.5135	0.3922	0.5142	0.0	0.5135	0.5135
No log	5	890	0.8761	0.125	1.8274	0.6719	0.4983	0.6733	0.0	0.6726	0.6719
0.0679	6	1068	0.7637	0.25	2.3717	0.6982	0.5558	0.6989	0.0	0.6996	0.6982
0.677	7	1246	0.7559	0.5	3.5464	0.7116	0.7000	0.7124	0.0	0.7124	0.7109
0.5869	8.0	1424	0.6202	1.0	5.9121	0.7550	0.7665	0.7557	0.0	0.7557	0.7550
0.4503	9.0	1602	0.7160	1.0	5.7949	0.7315	0.7220	0.7322	0.0	0.7322	0.7308
0.3411	10.0	1780	0.7409	1.0	5.8967	0.7521	0.7834	0.7536	0.0	0.7528	0.7521
0.2633	11.0	1958	0.8448	1.0	5.8588	0.7443	0.7886	0.7450	0.0	0.7450	0.7443
0.1718	12.0	2136	0.9531	1.0	5.7829	0.7479	0.7743	0.7486	0.0	0.7482	0.7479

Framework versions

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