ee7e6d358489e894b70dc6d388a48309

This model is a fine-tuned version of studio-ousia/mluke-base-lite on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 0.3713
• Data Size: 1.0
• Epoch Runtime: 59.8566
• Accuracy: 0.8965
• F1 Macro: 0.7657

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	1.1531	0	4.4756	0.1392	0.1209
No log	1	619	0.6623	0.0078	5.2800	0.7672	0.2894
No log	2	1238	0.5114	0.0156	5.7047	0.7672	0.2894
0.0134	3	1857	0.5262	0.0312	6.8084	0.8492	0.5263
0.0134	4	2476	0.3720	0.0625	8.4090	0.8787	0.5868
0.3224	5	3095	0.3131	0.125	12.1374	0.8896	0.6428
0.0257	6	3714	0.2965	0.25	19.3942	0.9040	0.6242
0.28	7	4333	0.2908	0.5	33.7691	0.8931	0.7616
0.2414	8.0	4952	0.2653	1.0	61.1200	0.9166	0.7512
0.2027	9.0	5571	0.2671	1.0	61.7350	0.9156	0.7645
0.1801	10.0	6190	0.2772	1.0	62.6450	0.9060	0.7749
0.166	11.0	6809	0.3143	1.0	60.1821	0.9093	0.7569
0.1185	12.0	7428	0.3713	1.0	59.8566	0.8965	0.7657

Framework versions

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