prompt-attack-minilm

This model is a fine-tuned version of sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.0527
• F1: 0.9866
• Precision: 1.0
• Recall: 0.9735
• Roc Auc: 0.9951

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: 3e-05
• train_batch_size: 32
• eval_batch_size: 64
• 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: 100
• num_epochs: 3
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	F1	Precision	Recall	Roc Auc
0.6324	0.1880	50	0.5153	0.6188	0.9583	0.4570	0.9265
0.1492	0.3759	100	0.1515	0.9420	0.9718	0.9139	0.9847
0.1063	0.5639	150	0.0759	0.9735	0.9735	0.9735	0.9922
0.0692	0.7519	200	0.0826	0.9695	0.9931	0.9470	0.9923
0.0534	0.9398	250	0.0726	0.9732	0.9864	0.9603	0.9938
0.0326	1.1278	300	0.0507	0.9766	0.9865	0.9669	0.9941
0.0274	1.3158	350	0.0505	0.9833	0.9932	0.9735	0.9943
0.0220	1.5038	400	0.0764	0.9763	1.0	0.9536	0.9939
0.0347	1.6917	450	0.0786	0.9728	1.0	0.9470	0.9945
0.0207	1.8797	500	0.0435	0.9833	0.9932	0.9735	0.9956
0.0129	2.0677	550	0.0723	0.9763	1.0	0.9536	0.9957
0.0154	2.2556	600	0.0500	0.9832	1.0	0.9669	0.9957
0.0034	2.4436	650	0.0581	0.9832	1.0	0.9669	0.9950
0.0089	2.6316	700	0.0527	0.9866	1.0	0.9735	0.9951
0.0171	2.8195	750	0.0508	0.9866	1.0	0.9735	0.9954
0.0249	3.0	798	0.0433	0.9866	1.0	0.9735	0.9957

Framework versions

• Transformers 5.17.0
• Pytorch 2.14.0+cu130
• Datasets 5.0.1
• Tokenizers 0.23.2

License

All rights reserved. No reuse, redistribution, or derivative works without explicit permission.