QA-DeBERTa-v3-large-diff-binary-2

saiteki-kai/QA-DeBERTa-v3-large-diff-binary-2

QA-DeBERTa-v3-large-diff-binary-2 is a text classification model from saiteki-kai. Use it when you need a label for a piece of text. It is set up for transformers. The card lists the license as mit.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

33% of all-time downloads

All-time downloads

Public

Parameters

439M

17.6 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors1.8 GB · 99%

At a glance

Task: Text Classification
Library: transformers
License: mit
Model type: deberta-v2
Access: Public
Created: Nov 28, 2025
Updated: Nov 29, 2025
SHA: fbcfe152

Try a prompt

Base models

microsoft/deberta-v3-large

Datasets

beavertails

Model card

From the Hugging Face model README

QA-DeBERTa-v3-large-diff-binary-2

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: 0.3409
Accuracy: 0.8609
Unsafe Precision: 0.8682
Unsafe Recall: 0.8842
Unsafe F1: 0.8761
Unsafe Fpr: 0.1684
Unsafe Aucpr: 0.9538
Safe Precision: 0.8512
Safe Recall: 0.8316
Safe F1: 0.8413
Safe Fpr: 0.1158
Safe Aucpr: 0.9184

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: 6e-06
train_batch_size: 64
eval_batch_size: 64
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
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
0.2998	0.2501	2114	0.3677	0.8446	0.9027	0.8078	0.8526	0.1093	0.9436	0.7870	0.8907	0.8356	0.1922	0.8961
0.3262	0.5001	4228	0.3278	0.8561	0.8786	0.8602	0.8693	0.1491	0.9495	0.8291	0.8509	0.8399	0.1398	0.9087
0.3019	0.7502	6342	0.3236	0.8588	0.8972	0.8429	0.8692	0.1211	0.9527	0.8168	0.8789	0.8467	0.1571	0.9155
0.3479	1.0002	8456	0.3215	0.8599	0.8690	0.8811	0.8750	0.1666	0.9531	0.8482	0.8334	0.8407	0.1189	0.9175
0.302	1.2503	10570	0.3221	0.8611	0.8839	0.8639	0.8738	0.1423	0.9536	0.8340	0.8577	0.8457	0.1361	0.9176
0.2663	1.5004	12684	0.3409	0.8609	0.8682	0.8842	0.8761	0.1684	0.9538	0.8512	0.8316	0.8413	0.1158	0.9184
0.2841	1.7504	14798	0.3223	0.8622	0.8772	0.8748	0.8760	0.1537	0.9551	0.8435	0.8463	0.8449	0.1252	0.9202
0.3074	2.0005	16912	0.3244	0.8632	0.8995	0.8490	0.8735	0.1190	0.9553	0.8230	0.8810	0.8510	0.1510	0.9182
0.3052	2.2505	19026	0.3200	0.8618	0.8833	0.8660	0.8746	0.1435	0.9546	0.8359	0.8565	0.8461	0.1340	0.9221
0.268	2.5006	21140	0.3192	0.8627	0.8876	0.8625	0.8748	0.1370	0.9550	0.8334	0.8630	0.8479	0.1375	0.9220

Framework versions

Transformers 4.57.3
Pytorch 2.7.1+cu118
Datasets 4.4.1
Tokenizers 0.22.1

Task: Text Classification
Library: transformers
Type: deberta-v2
License: mit
Created: Nov 28, 2025
Updated: Nov 29, 2025