5051cd00c8b9dca8c28cc0f52d87ca4a

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

• Loss: 0.6982
• Data Size: 1.0
• Epoch Runtime: 9.8176
• Mse: 0.6983
• Mae: 0.6328
• R2: 0.6876

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	Mse	Mae	R2
No log	0	0	8.4757	0	1.4144	8.4770	2.5072	-2.7921
No log	1	179	2.8427	0.0078	1.5485	2.8436	1.4332	-0.2721
No log	2	358	3.6737	0.0156	1.4753	3.6743	1.5435	-0.6436
No log	3	537	2.1996	0.0312	1.8633	2.2002	1.2260	0.0158
No log	4	716	1.0600	0.0625	2.1372	1.0604	0.8035	0.5256
No log	5	895	0.9174	0.125	2.6375	0.9178	0.8078	0.5894
0.1032	6	1074	0.9261	0.25	3.6814	0.9264	0.7233	0.5856
0.8936	7	1253	0.8148	0.5	5.8336	0.8153	0.7300	0.6353
0.7152	8.0	1432	0.7638	1.0	10.2491	0.7640	0.6858	0.6582
0.4503	9.0	1611	0.7959	1.0	10.1196	0.7962	0.6707	0.6438
0.3382	10.0	1790	0.7601	1.0	10.3974	0.7603	0.6570	0.6599
0.2725	11.0	1969	0.8345	1.0	10.6685	0.8348	0.6909	0.6266
0.2263	12.0	2148	0.7225	1.0	9.9498	0.7228	0.6456	0.6767
0.1834	13.0	2327	0.8246	1.0	9.8092	0.8247	0.6678	0.6311
0.1575	14.0	2506	0.7341	1.0	9.9655	0.7344	0.6461	0.6715
0.145	15.0	2685	0.7310	1.0	9.9397	0.7313	0.6549	0.6729
0.1208	16.0	2864	0.7148	1.0	10.2383	0.7150	0.6458	0.6802
0.1101	17.0	3043	0.6975	1.0	10.5535	0.6978	0.6334	0.6879
0.1091	18.0	3222	0.7020	1.0	9.7789	0.7022	0.6379	0.6859
0.0898	19.0	3401	0.6837	1.0	9.7935	0.6838	0.6256	0.6941
0.0947	20.0	3580	0.7048	1.0	9.7814	0.7050	0.6362	0.6846
0.0864	21.0	3759	0.6780	1.0	9.8035	0.6783	0.6181	0.6966
0.0737	22.0	3938	0.6902	1.0	9.8960	0.6904	0.6296	0.6911
0.0738	23.0	4117	0.6759	1.0	10.3986	0.6761	0.6088	0.6975
0.0621	24.0	4296	0.6782	1.0	10.0144	0.6784	0.6222	0.6965
0.0697	25.0	4475	0.7305	1.0	10.2331	0.7308	0.6370	0.6731
0.0686	26.0	4654	0.6889	1.0	9.8607	0.6892	0.6226	0.6917
0.0624	27.0	4833	0.6982	1.0	9.8176	0.6983	0.6328	0.6876

Framework versions

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