test_dir_model3

This model is a fine-tuned version of cardiffnlp/twitter-roberta-base-hate on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.2679
• Accuracy: 0.8783
• F1: 0.6997
• Precision: 0.8118
• Recall: 0.6915

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: 2e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH_FUSED with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: cosine
• num_epochs: 10
• label_smoothing_factor: 0.1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	F1	Precision	Recall
No log	1.0	35	1.1935	0.7304	0.3920	0.3732	0.4740
No log	2.0	70	1.2142	0.8348	0.6136	0.6065	0.6625
0.9237	3.0	105	1.3719	0.8783	0.7167	0.7802	0.7244
0.9237	4.0	140	1.5517	0.8696	0.7135	0.7685	0.7367
0.9237	5.0	175	1.9028	0.8870	0.7282	0.7874	0.7423
0.3387	6.0	210	2.0006	0.8696	0.6333	0.775	0.6708
0.3387	7.0	245	2.1684	0.8783	0.6997	0.8118	0.6915
0.3387	8.0	280	2.1672	0.8696	0.6958	0.7972	0.7038
0.119	9.0	315	2.2526	0.8783	0.6997	0.8118	0.6915
0.119	10.0	350	2.2679	0.8783	0.6997	0.8118	0.6915

Framework versions

• Transformers 4.57.3
• Pytorch 2.9.0+cu126
• Datasets 4.0.0
• Tokenizers 0.22.1