dadf6c3e0298d464ea5cf4489e4a63a7

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

• Loss: 0.6772
• Data Size: 1.0
• Epoch Runtime: 119.5790
• 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	1.1117	0	8.8250	0.1727	0.0982
No log	1	619	1.1463	0.0078	9.6626	0.2435	0.1627
No log	2	1238	0.5634	0.0156	11.1599	0.8212	0.4624
0.0161	3	1857	0.4317	0.0312	13.3897	0.8989	0.5971
0.0161	4	2476	0.3876	0.0625	17.5186	0.9071	0.6151
0.3822	5	3095	0.3616	0.125	25.1063	0.8888	0.6784
0.0423	6	3714	0.3648	0.25	38.4342	0.9014	0.6025
0.4253	7	4333	0.4048	0.5	66.5569	0.8949	0.5937
0.6797	8.0	4952	0.6772	1.0	120.5429	0.7672	0.2894
0.6497	9.0	5571	0.6772	1.0	119.5790	0.7672	0.2894

Framework versions

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