eb1f6a4cb140a32d27e2d672d9f1803c

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

• Loss: 0.7354
• Data Size: 1.0
• Epoch Runtime: 133.9749
• Accuracy: 0.9030
• F1 Macro: 0.7292

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	0.9168	0	11.1781	0.6408	0.3081
No log	1	619	0.6065	0.0078	12.2449	0.7810	0.3498
No log	2	1238	0.3482	0.0156	14.0456	0.8860	0.6609
0.0107	3	1857	0.3056	0.0312	17.4989	0.8983	0.6095
0.0107	4	2476	0.2784	0.0625	22.2274	0.9077	0.7226
0.3038	5	3095	0.2664	0.125	30.4617	0.9097	0.7282
0.0241	6	3714	0.2712	0.25	46.5622	0.9081	0.6443
0.2465	7	4333	0.2771	0.5	76.7011	0.9012	0.7526
0.2021	8.0	4952	0.2523	1.0	137.8459	0.9148	0.7587
0.1052	9.0	5571	0.3590	1.0	140.5121	0.9014	0.7337
0.0581	10.0	6190	0.4576	1.0	133.4485	0.9079	0.7236
0.0278	11.0	6809	0.5997	1.0	138.0240	0.8977	0.7100
0.02	12.0	7428	0.7354	1.0	133.9749	0.9030	0.7292

Framework versions

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