f0e34afffe85e814b4e41abc80263fca

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

• Loss: 0.3189
• Data Size: 1.0
• Epoch Runtime: 49.4083
• Accuracy: 0.9115
• F1 Macro: 0.7587

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	1.2039	0	4.1789	0.0601	0.0378
No log	1	619	0.6824	0.0078	4.7944	0.7672	0.2894
No log	2	1238	0.6033	0.0156	5.1378	0.7672	0.2894
0.0148	3	1857	0.3472	0.0312	5.9556	0.8904	0.5871
0.0148	4	2476	0.2996	0.0625	7.7942	0.9046	0.6062
0.3137	5	3095	0.2770	0.125	10.1123	0.9065	0.7305
0.0257	6	3714	0.2918	0.25	15.7152	0.9054	0.6093
0.2836	7	4333	0.3122	0.5	27.3986	0.8981	0.7566
0.2616	8.0	4952	0.2436	1.0	50.3288	0.9164	0.6985
0.2095	9.0	5571	0.3629	1.0	49.8399	0.8843	0.6892
0.2033	10.0	6190	0.2972	1.0	49.5511	0.9150	0.7817
0.2164	11.0	6809	0.2992	1.0	50.8395	0.9093	0.7611
0.1258	12.0	7428	0.3189	1.0	49.4083	0.9115	0.7587

Framework versions

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