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: 0.3184
Accuracy: 0.8622
Unsafe Precision: 0.8690
Unsafe Recall: 0.8859
Unsafe F1: 0.8773
Unsafe Fpr: 0.1676
Unsafe Aucpr: 0.9546
Safe Precision: 0.8533
Safe Recall: 0.8324
Safe F1: 0.8427
Safe Fpr: 0.1141
Safe Aucpr: 0.9201

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: 5e-06
train_batch_size: 64
eval_batch_size: 512
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 256
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
lr_scheduler_warmup_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
0.3322	1.0	2114	0.3267	0.8582	0.8761	0.8678	0.8720	0.1539	0.9505	0.8361	0.8461	0.8411	0.1322	0.9121
0.3217	2.0	4228	0.3184	0.8621	0.8689	0.8858	0.8773	0.1676	0.9546	0.8532	0.8324	0.8427	0.1142	0.9201
0.2961	3.0	6342	0.3166	0.8626	0.8939	0.8546	0.8738	0.1272	0.9557	0.8271	0.8728	0.8493	0.1454	0.9224
0.2987	4.0	8456	0.3153	0.8629	0.8866	0.8643	0.8753	0.1387	0.9559	0.8349	0.8613	0.8479	0.1357	0.9230
0.266	5.0	10570	0.3344	0.8606	0.8727	0.8776	0.8751	0.1606	0.9546	0.8453	0.8394	0.8423	0.1224	0.9225
0.2641	6.0	12684	0.3452	0.8607	0.8860	0.8604	0.8730	0.1389	0.9542	0.8310	0.8611	0.8458	0.1396	0.9216

Framework versions

Transformers 4.57.1
Pytorch 2.7.1+cu118
Datasets 4.4.1
Tokenizers 0.22.1

saiteki-kai/QA-DeBERTa-v3-large-binary-3

QA-DeBERTa-v3-large-binary-3

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

saiteki-kai/
QA-DeBERTa-v3-large-binary-3