german-sentiment-bert-finetuned-augmentation

This model is a fine-tuned version of oliverguhr/german-sentiment-bert on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.5289
• F1: 0.4075
• Roc Auc: 0.6477
• Accuracy: 0.3476

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: 2e-05
• train_batch_size: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_steps: 100
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Roc Auc	Accuracy
0.4252	1.0	141	0.4303	0.2285	0.5735	0.3244
0.3796	2.0	282	0.4097	0.2975	0.6081	0.3512
0.34	3.0	423	0.4223	0.2989	0.6109	0.3494
0.3083	4.0	564	0.4245	0.3105	0.6188	0.3529
0.2541	5.0	705	0.4319	0.3399	0.6179	0.3565
0.2404	6.0	846	0.4361	0.3412	0.6222	0.3583
0.2196	7.0	987	0.4547	0.3744	0.6344	0.3583
0.2299	8.0	1128	0.4542	0.3709	0.6334	0.3512
0.2	9.0	1269	0.4648	0.3502	0.6229	0.3601
0.1662	10.0	1410	0.4873	0.3746	0.6345	0.3440
0.1677	11.0	1551	0.4975	0.3920	0.6454	0.3601
0.1421	12.0	1692	0.5007	0.3844	0.6401	0.3494
0.1384	13.0	1833	0.5071	0.3836	0.6395	0.3529
0.1497	14.0	1974	0.5112	0.3837	0.6388	0.3672
0.1229	15.0	2115	0.5206	0.3950	0.6441	0.3458
0.1442	16.0	2256	0.5263	0.4015	0.6467	0.3494
0.1148	17.0	2397	0.5245	0.3996	0.6435	0.3547
0.1077	18.0	2538	0.5292	0.3977	0.6433	0.3369
0.1203	19.0	2679	0.5289	0.4051	0.6462	0.3422
0.1234	20.0	2820	0.5289	0.4075	0.6477	0.3476

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0