f7d07a20e62d248fb657955ab9e90472

This model is a fine-tuned version of google-bert/bert-base-multilingual-cased on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 0.3808
• Data Size: 1.0
• Epoch Runtime: 34.9468
• Accuracy: 0.8998
• F1 Macro: 0.7304

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	1.1057	0	2.9289	0.3119	0.2499
No log	1	619	0.6813	0.0078	3.3365	0.7672	0.2894
No log	2	1238	0.5677	0.0156	3.7019	0.7922	0.3895
0.0139	3	1857	0.3750	0.0312	4.5182	0.8689	0.5604
0.0139	4	2476	0.3505	0.0625	5.6733	0.8920	0.6616
0.3261	5	3095	0.3124	0.125	7.8129	0.8971	0.6883
0.0284	6	3714	0.2988	0.25	11.8296	0.9000	0.6011
0.2952	7	4333	0.3090	0.5	19.7089	0.8996	0.6004
0.2866	8.0	4952	0.2760	1.0	36.7967	0.9071	0.7321
0.2184	9.0	5571	0.2700	1.0	35.0163	0.9018	0.6043
0.2309	10.0	6190	0.3359	1.0	36.2578	0.9060	0.7501
0.1992	11.0	6809	0.3488	1.0	34.7964	0.9065	0.7394
0.1524	12.0	7428	0.3536	1.0	36.1305	0.9065	0.7468
0.1376	13.0	8047	0.3808	1.0	34.9468	0.8998	0.7304

Framework versions

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