AmirlyPhd/offensive_Bert_after_IEEE_reviwer

finetuned_bert

This model is a fine-tuned version of AmirlyPhd/final_final_bert_model_with_dynamic_resampling on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.8297
• Accuracy: 0.6973
• Precision: 0.6921
• Recall: 0.6973
• F1: 0.6940

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: 16
• eval_batch_size: 32
• seed: 42
• optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
• lr_scheduler_type: linear
• num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
0.2262	1.0	104	1.3268	0.6757	0.6999	0.6757	0.6815
0.4549	2.0	208	0.7178	0.6865	0.7011	0.6865	0.6910
0.2922	3.0	312	0.8297	0.6973	0.6921	0.6973	0.6940
0.2039	4.0	416	1.1670	0.6811	0.7035	0.6811	0.6867
0.147	5.0	520	1.3645	0.6595	0.6678	0.6595	0.6627
0.1551	6.0	624	1.6947	0.6324	0.6511	0.6324	0.6382
0.086	7.0	728	1.9108	0.6649	0.6744	0.6649	0.6684
0.0706	8.0	832	1.8700	0.6919	0.6891	0.6919	0.6903
0.0556	9.0	936	1.9929	0.6865	0.6932	0.6865	0.6891
0.016	10.0	1040	2.0539	0.6865	0.6932	0.6865	0.6891

Framework versions

• Transformers 4.51.3
• Pytorch 2.6.0+cu124
• Datasets 3.5.0
• Tokenizers 0.21.1