twitter-roberta-large-hate-latest-offensive-eval-kn

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

• Loss: 0.8543
• Accuracy: 0.7391
• Precision: 0.4215
• Recall: 0.3968
• F1: 0.4020

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: 1e-05
• train_batch_size: 16
• eval_batch_size: 128
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.9082	0.9968	157	0.8529	0.7286	0.3746	0.3354	0.3312
0.7593	2.0	315	0.7818	0.7393	0.5160	0.3778	0.3767
0.7264	2.9968	472	0.7640	0.7464	0.4450	0.3812	0.3861
0.6998	4.0	630	0.7941	0.7464	0.4461	0.4106	0.4218
0.5066	4.9968	787	0.8636	0.7518	0.4668	0.4156	0.4276
0.5164	6.0	945	0.8747	0.7482	0.4391	0.4342	0.4342
0.4098	6.9968	1102	0.9078	0.7446	0.4366	0.4324	0.4334
0.3556	8.0	1260	0.9286	0.7393	0.4279	0.4304	0.4282
0.3974	8.9968	1417	0.9444	0.7446	0.4434	0.4406	0.4411
0.318	9.9683	1570	0.9597	0.7411	0.4352	0.4370	0.4352

Framework versions

• Transformers 4.45.1
• Pytorch 2.4.0
• Datasets 3.0.1
• Tokenizers 0.20.0