28a8e53904780944e2889c99f7ae6b85

This model is a fine-tuned version of distilbert/distilgpt2 on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 0.5368
• Data Size: 1.0
• Epoch Runtime: 27.0331
• Accuracy: 0.8908
• F1 Macro: 0.7376

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	6.1676	0	3.2324	0.0633	0.0408
No log	1	619	1.7804	0.0078	3.5676	0.3488	0.2527
No log	2	1238	0.7792	0.0156	3.6281	0.7240	0.3474
0.0323	3	1857	0.5612	0.0312	4.1120	0.7991	0.4307
0.0323	4	2476	0.3725	0.0625	4.8251	0.8762	0.5798
0.343	5	3095	0.2935	0.125	6.5005	0.9012	0.7077
0.0258	6	3714	0.2748	0.25	9.3493	0.9083	0.7111
0.2595	7	4333	0.2718	0.5	15.6207	0.9014	0.7517
0.2256	8.0	4952	0.2458	1.0	28.5792	0.9152	0.7335
0.1807	9.0	5571	0.2761	1.0	29.0246	0.9044	0.7471
0.1603	10.0	6190	0.3406	1.0	27.0634	0.9060	0.7552
0.1266	11.0	6809	0.3590	1.0	27.1792	0.9014	0.7412
0.067	12.0	7428	0.5368	1.0	27.0331	0.8908	0.7376

Framework versions

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