distilbert-base-uncased__hate_speech_offensive__train-32-2

This model is a fine-tuned version of distilbert-base-uncased on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7136
• Accuracy: 0.679

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: 4
• eval_batch_size: 4
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 50
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1052	1.0	19	1.0726	0.45
1.0421	2.0	38	1.0225	0.5
0.9173	3.0	57	0.9164	0.6
0.6822	4.0	76	0.8251	0.7
0.4407	5.0	95	0.8908	0.5
0.2367	6.0	114	0.6772	0.75
0.1145	7.0	133	0.7792	0.65
0.0479	8.0	152	1.0657	0.6
0.0186	9.0	171	1.2228	0.65
0.0111	10.0	190	1.1100	0.6
0.0083	11.0	209	1.1991	0.65
0.0067	12.0	228	1.2654	0.65
0.0061	13.0	247	1.2837	0.65
0.0046	14.0	266	1.2860	0.6
0.0043	15.0	285	1.3160	0.65
0.0037	16.0	304	1.3323	0.65

Framework versions

• Transformers 4.15.0
• Pytorch 1.10.2+cu102
• Datasets 1.18.2
• Tokenizers 0.10.3