perspective-toxic-bert

This model is a fine-tuned version of unitary/toxic-bert on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.4656
• Tox Mse: 0.0239
• Tox Auc: 0.9631
• Dir Accuracy: 0.8345
• Dir F1 Macro: 0.5567
• Dir F1 Weighted: 0.8714

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: 32
• 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: linear
• lr_scheduler_warmup_steps: 0.06
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Tox Mse	Tox Auc	Dir Accuracy	Dir F1 Macro	Dir F1 Weighted
0.6404	1.0	997	0.4656	0.0239	0.9631	0.8345	0.5567	0.8714

Framework versions

• Transformers 5.7.0
• Pytorch 2.11.0+cu130
• Datasets 4.8.5
• Tokenizers 0.22.2