DeBERTa-QA-008

saiteki-kai/DeBERTa-QA-008

DeBERTa-QA-008 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

10% of all-time downloads

All-time downloads

Public

Parameters

435M

13.9 GB on disk

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors1.7 GB · 99%

At a glance

Task: Text Classification
Library: transformers
License: mit
Model type: deberta-v2
Access: Public
Created: Nov 15, 2025
Updated: Nov 15, 2025
SHA: 01c2fc4c

Try a prompt

Base models

microsoft/deberta-v3-large

Datasets

beavertails

Model card

From the Hugging Face model README

DeBERTa-QA-008

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: 0.7239
Accuracy: 0.0
Macro F1: 0.1258
Macro Precision: 0.0766
Macro Recall: 0.6798
Micro F1: 0.1575
Micro Precision: 0.0871
Micro Recall: 0.8240
Flagged/accuracy: 0.5565
Flagged/precision: 0.5565
Flagged/recall: 1.0
Flagged/f1: 0.7150
Flagged/aucpr: 0.7782
Flagged/fpr: 1.0000
Animal Abuse/accuracy: 0.9494
Animal Abuse/precision: 0.0140
Animal Abuse/recall: 0.0494
Animal Abuse/f1: 0.0219
Animal Abuse/fpr: 0.0402
Animal Abuse/threshold: 0.6039
Child Abuse/accuracy: 0.0959
Child Abuse/precision: 0.0057
Child Abuse/recall: 0.9309
Child Abuse/f1: 0.0113
Child Abuse/fpr: 0.9087
Child Abuse/threshold: 0.5675
Controversial Topics,politics/accuracy: 0.5387
Controversial Topics,politics/precision: 0.0357
Controversial Topics,politics/recall: 0.5402
Controversial Topics,politics/f1: 0.0670
Controversial Topics,politics/fpr: 0.4614
Controversial Topics,politics/threshold: 0.4634
Discrimination,stereotype,injustice/accuracy: 0.0802
Discrimination,stereotype,injustice/precision: 0.0795
Discrimination,stereotype,injustice/recall: 0.9994
Discrimination,stereotype,injustice/f1: 0.1473
Discrimination,stereotype,injustice/fpr: 0.9992
Discrimination,stereotype,injustice/threshold: 0.4727
Drug Abuse,weapons,banned Substance/accuracy: 0.0569
Drug Abuse,weapons,banned Substance/precision: 0.0563
Drug Abuse,weapons,banned Substance/recall: 0.9991
Drug Abuse,weapons,banned Substance/f1: 0.1066
Drug Abuse,weapons,banned Substance/fpr: 0.9993
Drug Abuse,weapons,banned Substance/threshold: 0.4268
Financial Crime,property Crime,theft/accuracy: 0.0983
Financial Crime,property Crime,theft/precision: 0.0974
Financial Crime,property Crime,theft/recall: 0.9993
Financial Crime,property Crime,theft/f1: 0.1775
Financial Crime,property Crime,theft/fpr: 0.9988
Financial Crime,property Crime,theft/threshold: 0.4730
Hate Speech,offensive Language/accuracy: 0.7592
Hate Speech,offensive Language/precision: 0.1539
Hate Speech,offensive Language/recall: 0.3754
Hate Speech,offensive Language/f1: 0.2183
Hate Speech,offensive Language/fpr: 0.2031
Hate Speech,offensive Language/threshold: 0.5011
Misinformation Regarding Ethics,laws And Safety/accuracy: 0.8773
Misinformation Regarding Ethics,laws And Safety/precision: 0.0135
Misinformation Regarding Ethics,laws And Safety/recall: 0.1259
Misinformation Regarding Ethics,laws And Safety/f1: 0.0243
Misinformation Regarding Ethics,laws And Safety/fpr: 0.1134
Misinformation Regarding Ethics,laws And Safety/threshold: 0.5562
Non Violent Unethical Behavior/accuracy: 0.1987
Non Violent Unethical Behavior/precision: 0.1987
Non Violent Unethical Behavior/recall: 0.9999
Non Violent Unethical Behavior/f1: 0.3315
Non Violent Unethical Behavior/fpr: 1.0000
Non Violent Unethical Behavior/threshold: 0.5315
Privacy Violation/accuracy: 0.6658
Privacy Violation/precision: 0.0674
Privacy Violation/recall: 0.4494
Privacy Violation/f1: 0.1171
Privacy Violation/fpr: 0.3230
Privacy Violation/threshold: 0.5319
Self Harm/accuracy: 0.0094
Self Harm/precision: 0.0068
Self Harm/recall: 0.9976
Self Harm/f1: 0.0136
Self Harm/fpr: 0.9974
Self Harm/threshold: 0.5465
Sexually Explicit,adult Content/accuracy: 0.2576
Sexually Explicit,adult Content/precision: 0.0247
Sexually Explicit,adult Content/recall: 0.7747
Sexually Explicit,adult Content/f1: 0.0478
Sexually Explicit,adult Content/fpr: 0.7552
Sexually Explicit,adult Content/threshold: 0.5467
Terrorism,organized Crime/accuracy: 0.7194
Terrorism,organized Crime/precision: 0.0117
Terrorism,organized Crime/recall: 0.4075
Terrorism,organized Crime/f1: 0.0227
Terrorism,organized Crime/fpr: 0.2781
Terrorism,organized Crime/threshold: 0.4249
Violence,aiding And Abetting,incitement/accuracy: 0.4442
Violence,aiding And Abetting,incitement/precision: 0.3072
Violence,aiding And Abetting,incitement/recall: 0.8679
Violence,aiding And Abetting,incitement/f1: 0.4538
Violence,aiding And Abetting,incitement/fpr: 0.7094
Violence,aiding And Abetting,incitement/threshold: 0.4801

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: 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_ratio: 0.1
num_epochs: 10

Training results

Training Loss	Epoch	Step	Validation Loss	Macro F1	Macro Precision	Macro Recall	Micro F1	Micro Precision	Micro Recall	Flagged/accuracy	Flagged/precision	Flagged/recall	Flagged/f1	Flagged/aucpr	Flagged/fpr	Animal Abuse/accuracy	Animal Abuse/precision	Animal Abuse/recall	Animal Abuse/f1	Animal Abuse/fpr	Animal Abuse/threshold	Child Abuse/accuracy	Child Abuse/precision	Child Abuse/recall	Child Abuse/f1	Child Abuse/fpr	Child Abuse/threshold	Controversial Topics,politics/accuracy	Controversial Topics,politics/precision	Controversial Topics,politics/recall	Controversial Topics,politics/f1	Controversial Topics,politics/fpr	Controversial Topics,politics/threshold	Discrimination,stereotype,injustice/accuracy	Discrimination,stereotype,injustice/precision	Discrimination,stereotype,injustice/recall	Discrimination,stereotype,injustice/f1	Discrimination,stereotype,injustice/fpr	Discrimination,stereotype,injustice/threshold	Drug Abuse,weapons,banned Substance/accuracy	Drug Abuse,weapons,banned Substance/precision	Drug Abuse,weapons,banned Substance/recall	Drug Abuse,weapons,banned Substance/f1	Drug Abuse,weapons,banned Substance/fpr	Drug Abuse,weapons,banned Substance/threshold	Financial Crime,property Crime,theft/accuracy	Financial Crime,property Crime,theft/precision	Financial Crime,property Crime,theft/recall	Financial Crime,property Crime,theft/f1	Financial Crime,property Crime,theft/fpr	Financial Crime,property Crime,theft/threshold	Hate Speech,offensive Language/accuracy	Hate Speech,offensive Language/precision	Hate Speech,offensive Language/recall	Hate Speech,offensive Language/f1	Hate Speech,offensive Language/fpr	Hate Speech,offensive Language/threshold	Misinformation Regarding Ethics,laws And Safety/accuracy	Misinformation Regarding Ethics,laws And Safety/precision	Misinformation Regarding Ethics,laws And Safety/recall	Misinformation Regarding Ethics,laws And Safety/f1	Misinformation Regarding Ethics,laws And Safety/fpr	Misinformation Regarding Ethics,laws And Safety/threshold	Non Violent Unethical Behavior/accuracy	Non Violent Unethical Behavior/precision	Non Violent Unethical Behavior/recall	Non Violent Unethical Behavior/f1	Non Violent Unethical Behavior/fpr	Non Violent Unethical Behavior/threshold	Privacy Violation/accuracy	Privacy Violation/precision	Privacy Violation/recall	Privacy Violation/f1	Privacy Violation/fpr	Privacy Violation/threshold	Self Harm/accuracy	Self Harm/precision	Self Harm/recall	Self Harm/f1	Self Harm/fpr	Self Harm/threshold	Sexually Explicit,adult Content/accuracy	Sexually Explicit,adult Content/precision	Sexually Explicit,adult Content/recall	Sexually Explicit,adult Content/f1	Sexually Explicit,adult Content/fpr	Sexually Explicit,adult Content/threshold	Terrorism,organized Crime/accuracy	Terrorism,organized Crime/precision	Terrorism,organized Crime/recall	Terrorism,organized Crime/f1	Terrorism,organized Crime/fpr	Terrorism,organized Crime/threshold	Violence,aiding And Abetting,incitement/accuracy	Violence,aiding And Abetting,incitement/precision	Violence,aiding And Abetting,incitement/recall	Violence,aiding And Abetting,incitement/f1	Violence,aiding And Abetting,incitement/fpr	Violence,aiding And Abetting,incitement/threshold
0.7331	0.0101	85	0.7334	0.1262	0.0772	0.7236	0.1469	0.0806	0.8231	0.5565	0.5565	1.0	0.7150	0.7782	1.0000	0.8739	0.0145	0.1497	0.0265	0.1177	0.6072	0.0761	0.0056	0.9399	0.0111	0.9288	0.5742	0.1965	0.0315	0.8485	0.0608	0.8241	0.4671	0.0797	0.0795	0.9996	0.1473	0.9997	0.4796	0.0569	0.0563	0.9994	0.1066	0.9994	0.4292	0.0985	0.0974	0.9991	0.1775	0.9986	0.4774	0.7697	0.1607	0.3724	0.2246	0.1913	0.5084	0.0758	0.0124	0.9535	0.0245	0.9350	0.5407	0.1987	0.1987	0.9999	0.3315	1.0000	0.5383	0.7184	0.0740	0.4090	0.1253	0.2656	0.5344	0.0173	0.0068	0.9805	0.0134	0.9894	0.5521	0.6057	0.0253	0.4098	0.0477	0.3894	0.5571	0.8362	0.0118	0.2349	0.0224	0.1590	0.4292	0.4519	0.3056	0.8337	0.4473	0.6865	0.4840
0.7294	0.0201	170	0.7303	0.1263	0.0769	0.7095	0.1521	0.0837	0.8304	0.5565	0.5565	1.0	0.7150	0.7782	1.0000	0.7397	0.0142	0.3169	0.0271	0.2554	0.6053	0.0983	0.0057	0.9309	0.0113	0.9064	0.5722	0.2398	0.0319	0.8100	0.0613	0.7782	0.4659	0.0800	0.0795	0.9996	0.1473	0.9995	0.4776	0.0569	0.0563	0.9991	0.1066	0.9994	0.4282	0.0987	0.0974	0.9988	0.1774	0.9984	0.4763	0.7585	0.1575	0.3899	0.2243	0.2052	0.5061	0.8811	0.0136	0.1231	0.0246	0.1095	0.5607	0.1987	0.1987	0.9999	0.3315	1.0000	0.5363	0.7129	0.0724	0.4076	0.1229	0.2712	0.5334	0.0120	0.0068	0.9927	0.0135	0.9948	0.5501	0.2199	0.0248	0.8210	0.0482	0.7949	0.5501	0.7963	0.0117	0.2931	0.0225	0.1997	0.4278	0.4470	0.3059	0.8501	0.4499	0.6992	0.4827
0.7196	0.0302	255	0.7239	0.1259	0.0767	0.6844	0.1573	0.0869	0.8265	0.5565	0.5565	1.0	0.7150	0.7782	1.0000	0.9491	0.0151	0.0538	0.0236	0.0405	0.6039	0.0957	0.0057	0.9339	0.0113	0.9089	0.5675	0.4720	0.0364	0.6374	0.0689	0.5332	0.4632	0.0802	0.0795	0.9994	0.1473	0.9992	0.4727	0.0569	0.0563	0.9991	0.1066	0.9993	0.4268	0.0983	0.0974	0.9997	0.1775	0.9989	0.4730	0.7592	0.1540	0.3756	0.2184	0.2030	0.5011	0.8891	0.0133	0.1108	0.0237	0.1013	0.5566	0.1987	0.1987	0.9999	0.3315	1.0000	0.5315	0.6660	0.0670	0.4461	0.1164	0.3226	0.5319	0.0094	0.0068	0.9976	0.0136	0.9974	0.5465	0.2744	0.0244	0.7491	0.0474	0.7373	0.5470	0.7197	0.0118	0.4116	0.0230	0.2778	0.4249	0.4440	0.3072	0.8684	0.4538	0.7098	0.4801
0.6887	0.0402	340	0.6877	0.1248	0.0758	0.6926	0.1533	0.0845	0.8270	0.5565	0.5565	1.0	0.7150	0.7782	1.0000	0.2318	0.0119	0.8081	0.0235	0.7748	0.5429	0.3783	0.0059	0.6637	0.0117	0.6233	0.5306	0.8377	0.0542	0.2611	0.0897	0.1441	0.4477	0.2248	0.0831	0.8722	0.1518	0.8311	0.4458	0.0637	0.0564	0.9941	0.1068	0.9919	0.4407	0.0973	0.0973	0.9998	0.1774	1.0000	0.4431	0.7222	0.1349	0.3884	0.2003	0.2449	0.4805	0.8991	0.0142	0.1067	0.0251	0.0912	0.5297	0.2006	0.1989	0.9982	0.3317	0.9973	0.4763	0.8499	0.0849	0.2090	0.1208	0.1169	0.5353	0.0123	0.0068	0.9951	0.0136	0.9945	0.5324	0.0251	0.0241	0.9993	0.0470	0.9989	0.5059	0.5719	0.0086	0.4574	0.0168	0.4272	0.4111	0.3377	0.2793	0.9428	0.4310	0.8816	0.4771
0.5072	0.0503	425	0.4784	0.1310	0.0866	0.6731	0.1442	0.0798	0.7528	0.5565	0.5565	1.0	0.7150	0.7782	1.0000	0.2757	0.0120	0.7631	0.0235	0.7300	0.3522	0.8275	0.0058	0.1772	0.0112	0.1689	0.3442	0.0308	0.0306	0.9995	0.0594	0.9999	0.2870	0.0802	0.0795	0.9985	0.1472	0.9991	0.3082	0.0975	0.0565	0.9572	0.1067	0.9538	0.3666	0.0973	0.0973	0.9998	0.1774	1.0000	0.2659	0.8394	0.2227	0.3184	0.2621	0.1093	0.3585	0.6142	0.0127	0.3995	0.0246	0.3831	0.2751	0.5041	0.2375	0.6768	0.3517	0.5388	0.3585	0.8900	0.1184	0.1908	0.1461	0.0737	0.4426	0.0543	0.0069	0.9659	0.0137	0.9520	0.3141	0.9045	0.0434	0.1410	0.0663	0.0767	0.3868	0.0629	0.0080	0.9397	0.0158	0.9441	0.3425	0.3635	0.2815	0.8967	0.4284	0.8297	0.4600

Framework versions

Transformers 4.57.1
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 15, 2025
Updated: Nov 15, 2025