50bcb1f851acf41d5683c7ce784f6ced

This model is a fine-tuned version of FacebookAI/xlm-roberta-large-finetuned-conll03-german on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 0.6807
• Data Size: 1.0
• Epoch Runtime: 105.8297
• Accuracy: 0.7672
• F1 Macro: 0.2894

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	0.9535	0	7.6883	0.7275	0.3101
No log	1	619	0.6722	0.0078	8.5303	0.7672	0.2894
No log	2	1238	0.7000	0.0156	10.0849	0.7672	0.2894
0.0162	3	1857	0.6645	0.0312	12.1809	0.7672	0.2894
0.0162	4	2476	0.6171	0.0625	15.7309	0.7685	0.2951
0.4647	5	3095	0.5582	0.125	22.2666	0.8421	0.5295
0.0641	6	3714	0.6925	0.25	34.6744	0.7672	0.2894
0.6706	7	4333	0.6799	0.5	59.1942	0.7672	0.2894
0.6773	8.0	4952	0.6804	1.0	106.9499	0.7672	0.2894
0.6494	9.0	5571	0.6807	1.0	105.8297	0.7672	0.2894

Framework versions

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