roberta-aigt-finetuning-v4

This model is a fine-tuned version of FacebookAI/roberta-base on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 0.6292
• Accuracy: 0.9587
• Precision Human@0.5: 0.8879
• Recall Human@0.5: 0.9456
• F1 Human@0.5: 0.9159
• Roc Auc: 0.9879
• Pr Auc: 0.9648
• Tpr@fpr1%: 0.7962

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: 64
• seed: 42
• gradient_accumulation_steps: 4
• total_train_batch_size: 64
• 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: cosine
• lr_scheduler_warmup_steps: 847
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision Human@0.5	Recall Human@0.5	F1 Human@0.5	Roc Auc	Pr Auc	Tpr@fpr1%
1.4794	0.3396	800	0.5455	0.8966	0.7494	0.8497	0.7964	0.9545	0.8750	0.4604
0.7314	0.6793	1600	0.6898	0.9061	0.7360	0.9444	0.8273	0.9738	0.9400	0.7191
0.2246	1.0187	2400	0.4291	0.9609	0.9319	0.9018	0.9166	0.9857	0.9666	0.8445
0.2752	1.3583	3200	0.7944	0.9323	0.8498	0.8694	0.8595	0.9712	0.9269	0.6334
0.2674	1.6979	4000	0.5984	0.9528	0.9220	0.8759	0.8984	0.9827	0.9527	0.7568
0.0919	2.0374	4800	0.6229	0.9585	0.8990	0.9302	0.9143	0.9862	0.9608	0.7893
0.0510	2.3770	5600	0.6230	0.9624	0.9302	0.9104	0.9202	0.9851	0.9646	0.8308
0.0655	2.7166	6400	0.6292	0.9587	0.8879	0.9456	0.9159	0.9879	0.9648	0.7962

Framework versions

• Transformers 5.0.0
• Pytorch 2.10.0+cu128
• Datasets 4.8.3
• Tokenizers 0.22.2