Modelpublic

saiteki-kai/QA-DeBERTa-v3-large-binary-focal

sourceHugging Facemitupdated 10mo agoView on Hugging Face

0likes15downloads

Model Card

QA-DeBERTa-v3-large-binary-focal

This model is a fine-tuned version of microsoft/deberta-v3-large on the saiteki-kai/Beavertails-it dataset. It achieves the following results on the evaluation set:

—Loss: 0.0603
—Accuracy: 0.8625
—Unsafe Precision: 0.8786
—Unsafe Recall: 0.8735
—Unsafe F1: 0.8761
—Unsafe Fpr: 0.1514
—Unsafe Aucpr: 0.9564
—Safe Precision: 0.8425
—Safe Recall: 0.8486
—Safe F1: 0.8455
—Safe Fpr: 0.1265
—Safe Aucpr: 0.9242

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: 9e-06
—trainbatchsize: 64
—evalbatchsize: 512
—seed: 42
—optimizer: Use OptimizerNames.ADAMWTORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizerargs=No additional optimizer arguments
—lrschedulertype: linear
—lrschedulerwarmup_steps: 1000
—num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Unsafe Precision	Unsafe Recall	Unsafe F1	Unsafe Fpr	Unsafe Aucpr	Safe Precision	Safe Recall	Safe F1	Safe Fpr	Safe Aucpr
0.0644	1.0	8454	0.0623	0.8604	0.8730	0.8766	0.8748	0.1600	0.9540	0.8444	0.8400	0.8422	0.1234	0.9189
0.0534	2.0	16908	0.0603	0.8625	0.8786	0.8736	0.8761	0.1515	0.9564	0.8425	0.8485	0.8455	0.1264	0.9242
0.0535	3.0	25362	0.0619	0.8615	0.8774	0.8730	0.8752	0.1530	0.9564	0.8417	0.8470	0.8443	0.1270	0.9252
0.0502	4.0	33816	0.0679	0.8584	0.8676	0.8797	0.8736	0.1684	0.9554	0.8464	0.8316	0.8389	0.1203	0.9252

Framework versions

—Transformers 4.57.1
—Pytorch 2.7.1+cu118
—Datasets 4.4.1
—Tokenizers 0.22.1