03d04ba1b5d43af630ff215e835fbf50

This model is a fine-tuned version of openai-community/gpt2 on the contemmcm/hate-speech-and-offensive-language dataset. It achieves the following results on the evaluation set:

• Loss: 0.5183
• Data Size: 1.0
• Epoch Runtime: 46.7536
• Accuracy: 0.8906
• F1 Macro: 0.7387

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	10.1280	0	4.8221	0.0645	0.0419
No log	1	619	3.4455	0.0078	5.1294	0.3466	0.2434
No log	2	1238	0.8010	0.0156	5.4392	0.6611	0.3691
0.0414	3	1857	0.5517	0.0312	6.4011	0.8070	0.4434
0.0414	4	2476	0.3677	0.0625	7.5354	0.8815	0.5874
0.3465	5	3095	0.2942	0.125	10.3462	0.8967	0.6938
0.0256	6	3714	0.2644	0.25	15.4469	0.9091	0.6863
0.2824	7	4333	0.2731	0.5	25.7869	0.8983	0.7601
0.2317	8.0	4952	0.2417	1.0	47.4210	0.9140	0.7558
0.1718	9.0	5571	0.2719	1.0	46.4659	0.9050	0.7600
0.1618	10.0	6190	0.3005	1.0	46.9503	0.9107	0.7604
0.1173	11.0	6809	0.3832	1.0	46.9982	0.9034	0.7446
0.0701	12.0	7428	0.5183	1.0	46.7536	0.8906	0.7387

Framework versions

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