f67ea2ac8487bb64cd44a7b4efe7d7e8

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

• Loss: 0.3215
• Data Size: 1.0
• Epoch Runtime: 9.5941
• Accuracy: 0.9375
• F1 Macro: 0.9378

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
No log	0	0	1.8343	0	0.7751	0.1562	0.1086
No log	1	170	1.8710	0.0078	1.1270	0.1354	0.0932
No log	2	340	1.8197	0.0156	1.0794	0.1792	0.1015
No log	3	510	1.5744	0.0312	1.3060	0.4042	0.2039
No log	4	680	0.9092	0.0625	1.6204	0.7542	0.6249
0.0733	5	850	0.4362	0.125	2.3236	0.8562	0.7235
0.0733	6	1020	0.3001	0.25	3.2927	0.9125	0.7626
0.3652	7	1190	0.2181	0.5	5.3010	0.9417	0.9290
0.2682	8.0	1360	0.2213	1.0	9.5133	0.9396	0.9173
0.1532	9.0	1530	0.2764	1.0	9.6768	0.9354	0.9326
0.0988	10.0	1700	0.3746	1.0	9.9725	0.9167	0.9059
0.0749	11.0	1870	0.3215	1.0	9.5941	0.9375	0.9378

Framework versions

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