26a587d1023dc16d7628058bdddbef54

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

• Loss: 0.8241
• Data Size: 1.0
• Epoch Runtime: 9.9129
• Accuracy: 0.7599
• F1 Macro: 0.7966
• Rouge1: 0.7599
• Rouge2: 0.0
• Rougel: 0.7599
• Rougelsum: 0.7592

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	1.5718	0	1.3189	0.2216	0.0931	0.2209	0.0	0.2216	0.2216
No log	1	178	1.3865	0.0078	1.6198	0.4304	0.2274	0.4304	0.0	0.4304	0.4297
No log	2	356	1.2159	0.0156	1.4881	0.5774	0.3914	0.5781	0.0	0.5781	0.5774
No log	3	534	1.0144	0.0312	1.7539	0.5661	0.4454	0.5661	0.0	0.5668	0.5668
No log	4	712	0.8029	0.0625	2.2474	0.6903	0.5397	0.6918	0.0	0.6918	0.6903
No log	5	890	0.8296	0.125	2.8198	0.6683	0.5100	0.6690	0.0	0.6690	0.6683
0.054	6	1068	0.6859	0.25	3.9845	0.7237	0.6987	0.7251	0.0	0.7237	0.7237
0.591	7	1246	0.6334	0.5	5.9675	0.75	0.7722	0.75	0.0	0.7507	0.75
0.4976	8.0	1424	0.6577	1.0	10.3981	0.7457	0.7614	0.7457	0.0	0.7468	0.7457
0.3702	9.0	1602	0.6680	1.0	10.3138	0.7585	0.7861	0.7592	0.0	0.7592	0.7585
0.2545	10.0	1780	0.9169	1.0	10.0893	0.7401	0.7762	0.7401	0.0	0.7408	0.7401
0.1905	11.0	1958	0.8241	1.0	9.9129	0.7599	0.7966	0.7599	0.0	0.7599	0.7592

Framework versions

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