# deberta-v3-base-prompt-injection-v2 by Protect AI Source: https://savrn.com/models/deberta-v3-base-prompt-injection-v2 Markdown alternate of the page above; the site index is https://savrn.com/llms.txt --- ## Runs On What it takes to serve deberta-v3-base-prompt-injection-v2 (184M parameters): the memory its weights need at each precision, and the cheapest way to rent enough data-center GPUs to hold them. | Precision | Weights | Memory needed | Cheapest setup | Per hour | Also fits | | --- | --- | --- | --- | --- | --- | | 16-bit | 0.4 GB | 0.4 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 8-bit | 0.2 GB | 0.2 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | | 4-bit | 0.1 GB | 0.1 GB | 1x [MI300X](https://savrn.com/ai-index/pricing/gpus/mi300x) (192 GB) Vultr | $1.85 | [1x H100](https://savrn.com/ai-index/pricing/gpus/h100) $1.99 · [1x MI325X](https://savrn.com/ai-index/pricing/gpus/mi325x) $2.00 | Memory is the weights at that precision plus 20% for the runtime and a short context; a long context needs more. Prices are the lowest on-demand hourly rates in the [SAVRN Index](https://savrn.com/ai-index/pricing/gpus), read Oct 22, 2026. [deberta-v3-base-prompt-injection-v2 on every accelerator the SAVRN Index prices, at every precision](https://savrn.com/models/deberta-v3-base-prompt-injection-v2/gpus) ## SAVRN's Notes on deberta-v3-base-prompt-injection-v2 Put this one in front of the generator, not behind it. Protect AI trained deberta-v3-base-prompt-injection-v2 to sort each input into benign, label 0, or injection detected, label 1. It needs 0.4 GB at 16-bit, 0.2 GB at 8-bit and 0.1 GB at 4-bit, so on the cheapest slot we price, one 192 GB MI300X at $1.85 an hour, it shares the card with the model it guards. The deployment is shaped by 512 tokens of context: anything longer must be windowed first, and that logic is yours to build. Apache 2.0 permits commercial use inside a paid product and redistribution with the notices kept. Check the lineage, listed in full: a fine-tuned version of microsoft/deberta-v3-base, trained on seven datasets including Harelix/Prompt-Injection-Mixed-Techniques-2024, jackhhao/jailbreak-classification and OpenSafetyLab/Salad-Data. Compare those against the traffic you actually receive. Released April 20, 2024, files updated July 9, 2026, so pull the current set, safetensors or ONNX. SAVRN Research, 2026-09-18 ## Model Card By Protect AI, published under apache-2.0, revision 90c9989b1a34. [!WARNING] THIS PROJECT HAS BEEN ARCHIVED. This project and its associated code on [GitHub](https://github.com/protectai/llm-guard) are no longer under active development or maintained. ### Model Card for deberta-v3-base-prompt-injection-v2 This model is a fine-tuned version of [microsoft/deberta-v3-base](https://savrn.com/models/deberta-v3-base) specifically developed to detect and classify prompt injection attacks which can manipulate language models into producing unintended outputs. ### Introduction Prompt injection attacks manipulate language models by inserting or altering prompts to trigger harmful or unintended responses. The deberta-v3-base-prompt-injection-v2 model is designed to enhance security in language model applications by detecting these malicious interventions. ### Model Details - Fine-tuned by: Protect AI - Model type: deberta-v3-base - Language(s) (NLP): English - License: Apache License 2.0 - Finetuned from model: [microsoft/deberta-v3-base](https://savrn.com/models/deberta-v3-base) ### Intended Uses This model classifies inputs into benign (0) and injection-detected (1). ### Limitations deberta-v3-base-prompt-injection-v2 is highly accurate in identifying prompt injections in English. It does not detect jailbreak attacks or handle non-English prompts, which may limit its applicability in diverse linguistic environments or against advanced adversarial techniques. [Read the full model card (544 words)](https://savrn.com/models/deberta-v3-base-prompt-injection-v2/card) ## Configuration Architecture DebertaV2ForSequenceClassification Context length (tokens) 512 Layers 12 Hidden size 768 Feed-forward size 3,072 Attention heads 12 Vocabulary size 128,100 Stored precision float32 Model type deberta-v2 ## Identity and Version Repository protectai/deberta-v3-base-prompt-injection-v2 Publisher Protect AI Task Text classification Modality Text Library transformers Parameters 184M parameters Languages en Revision 90c9989b1a342275dd0d1a95aad283c04e075671 First published 2024-04-20 Last updated 2026-07-09 ## Files and Weights 19 files, 1.5 GB in total. The weights are 3 files totalling 1.5 GB in bin, onnx, safetensors. Weights3 files · 1.5 GB Configuration6 files · 3.3 KB Tokenizer4 files · 17.3 MB Documentation2 files · 16.5 KB Other3 files · 4.9 MB Repository1 file · 1.5 KB Every file | File | Type | Size | SHA-256 | | --- | --- | --- | --- | | model.safetensors | Weights | 737.7 MB | 6521cb8d0ac0 | | onnx/model.onnx | Weights | 738.6 MB | f0ea7f239f76 | | training_args.bin | Weights | 5.0 KB | 850c358d86ce | | added_tokens.json | Configuration | 23 B | — | | config.json | Configuration | 994 B | — | | onnx/added_tokens.json | Configuration | 23 B | — | | onnx/config.json | Configuration | 1.0 KB | — | | onnx/special_tokens_map.json | Configuration | 970 B | — | | special_tokens_map.json | Configuration | 286 B | — | | LICENSE | Documentation | 10.2 KB | — | | README.md | Documentation | 6.3 KB | — | | deberta-v3-base-prompt-injection-v2_emissions.csv | Other | 810 B | — | | onnx/spm.model | Other | 2.5 MB | c679fbf93643 | | spm.model | Other | 2.5 MB | c679fbf93643 | | .gitattributes | Repository | 1.5 KB | — | | onnx/tokenizer.json | Tokenizer | 8.6 MB | — | | onnx/tokenizer_config.json | Tokenizer | 1.4 KB | — | | tokenizer.json | Tokenizer | 8.7 MB | — | | tokenizer_config.json | Tokenizer | 1.3 KB | — | ## License and Download License apache-2.0 Access Open weights, no gate Download size 1.5 GB [Download from Protect AI](https://huggingface.co/protectai/deberta-v3-base-prompt-injection-v2) Released by Protect AI through its official repository on Hugging Face. [Read the license](https://www.apache.org/licenses/LICENSE-2.0). ## Built From - Derived from [microsoft/deberta-v3-base](https://savrn.com/models/deberta-v3-base) - Quantized from [microsoft/deberta-v3-base](https://savrn.com/models/deberta-v3-base) - Trained on (disclosed) Harelix/Prompt-Injection-Mixed-Techniques-2024 - Trained on (disclosed) HuggingFaceH4/grok-conversation-harmless - Trained on (disclosed) OpenSafetyLab/Salad-Data - Trained on (disclosed) VMware/open-instruct - Trained on (disclosed) alespalla/chatbot_instruction_prompts - Trained on (disclosed) jackhhao/jailbreak-classification - Trained on (disclosed) natolambert/xstest-v2-copy ## Memory Requirements | Precision | Weights in memory | | --- | --- | | As published | 1.5 GB | | 16-bit | 0.4 GB | | 8-bit | 0.2 GB | | 4-bit | 0.1 GB | Weights only, from the published parameter count; the key-value cache and runtime add to this. ## Questions About deberta-v3-base-prompt-injection-v2 ### How much GPU memory does deberta-v3-base-prompt-injection-v2 need? About 0.4 GB at 16-bit and 0.1 GB at 4-bit: the weights (184M parameters) plus a working margin. A long context needs more. ### What is the cheapest GPU to run deberta-v3-base-prompt-injection-v2 on? At 16-bit, 1x MI300X from $1.85 an hour; at 4-bit, 1x MI300X from $1.85 an hour, at the lowest on-demand prices the SAVRN Index lists. ### Can I use deberta-v3-base-prompt-injection-v2 commercially? Yes. deberta-v3-base-prompt-injection-v2 is released under Apache License 2.0. The Apache License 2.0 is a permissive open-source license. It permits commercial use, modification and redistribution. It requires keeping the license and copyright notices and any NOTICE file, stating significant changes, and it includes an express patent grant from contributors. ### What is deberta-v3-base-prompt-injection-v2's context length? 512 tokens, from the maximum position embeddings in its published configuration. ## Similar Models Model · Text classification ### [DEBATE-kor-base](https://savrn.com/models/debate-kor-base) [Jong Rock Jeong](https://savrn.com/model-publishers/jongrock17) DEBATE-kor-base is a Korean-adapted Political DEBATE model for binary natural language inference (NLI) on political text. The model is initialized from mlburnham/PoliticalDEBATEDeBERTabasev1.1, the original DeBERTa-based Political DEBATE checkpoint, and subsequently fine-tuned on jongrock17/PolNLI-kor, a Korean translation and adaptation of PolNLI. Political DEBATE DeBERTa-base → PolNLI-kor → DEBATE-kor-base Unlike the PolNLI-kor-RoBERTa model family, which starts from Korean-pretrained KLUE-RoBERTa encoders, DEBATE-kor directly adapts the original Political DEBATE checkpoint to Korean political NLI. DEBATE-kor formulates NLI as a binary classification problem. notentailment combines… Open weights 184M parameters 512 tokens transformers [View model](https://savrn.com/models/debate-kor-base) Model · Text classification ### [DeBERTa-v3-base-mnli-fever-docnli-ling-2c](https://savrn.com/models/deberta-v3-base-mnli-fever-docnli-ling-2c) [Moritz Borrett-Laurer (formerly Laurer)](https://savrn.com/model-publishers/moritzlaurer) This model was trained on 1.279.665 hypothesis-premise pairs from 8 NLI datasets: MultiNLI, Fever-NLI, LingNLI and DocNLI (which includes ANLI, QNLI, DUC, CNN/DailyMail, Curation). It is the only model in the model hub trained on 8 NLI datasets, including DocNLI with very long texts to learn long range reasoning. Note that the model was trained on binary NLI to predict either "entailment" or "not-entailment". The DocNLI merges the classes "neural" and "contradiction" into "not-entailment" to enable the inclusion of the DocNLI dataset. The base model is DeBERTa-v3-base from Microsoft. The v3 variant of DeBERTa substantially outperforms previous versions of the model by including a different… Open weights mit 184M parameters 512 tokens transformers [View model](https://savrn.com/models/deberta-v3-base-mnli-fever-docnli-ling-2c) Model · Text classification ### [ai-agent-sme-sentiment-mbert](https://savrn.com/models/ai-agent-sme-sentiment-mbert) [Atsushi Hatakeyama](https://savrn.com/model-publishers/atsushihatake) This is a three-class Japanese/English sentiment classifier for short hospitality reviews, fine-tuned from google-bert/bert-base-multilingual-cased. It categorizes each review as negative, neutral, or positive. The model predicts one overall label for a short Japanese or English hospitality It is intended for research, prototyping, and human-reviewed analytics. It must not be used as the sole basis for consequential business, employment, moderation, or customer-service decisions. Applications using this model should allow operators to review and correct its predictions. The model was fine-tuned on 480 synthetic bilingual reviews: 240 Japanese and 240 English, balanced across the three… Open weights apache-2.0 178M parameters 512 tokens transformers [View model](https://savrn.com/models/ai-agent-sme-sentiment-mbert) Model · Text classification ### [mbert-sentimiento-multilingue](https://savrn.com/models/mbert-sentimiento-multilingue) [Juan Diego Iraheta Rubio](https://savrn.com/model-publishers/jdiraheta) This is the model card of a transformers model that has been pushed on the Hub. Users (both direct and downstream) should be made aware of the risks, biases and limitations of the model. Use the code below to get started with the model. Carbon emissions can be estimated using the Machine Learning Impact calculator presented in Lacoste et al. (2019). Open weights 178M parameters 512 tokens transformers [View model](https://savrn.com/models/mbert-sentimiento-multilingue) Model · Text classification ### [bert-base-multilingual-uncased-sentiment](https://savrn.com/models/bert-base-multilingual-uncased-sentiment) [NLP Town](https://savrn.com/model-publishers/nlptown) Visit the NLP Town website for an updated version of this model, with a 40% error reduction on product reviews. This is a bert-base-multilingual-uncased model finetuned for sentiment analysis on product reviews in six languages: English, Dutch, German, French, Spanish, and Italian. It predicts the sentiment of the review as a number of stars (between 1 and 5). This model is intended for direct use as a sentiment analysis model for product reviews in any of the six languages above or for further finetuning on related sentiment analysis tasks. Here is the number of product reviews we used for finetuning the model: The fine-tuned model obtained the following accuracy on 5,000 held-out product… Open weights mit 167M parameters 512 tokens transformers [View model](https://savrn.com/models/bert-base-multilingual-uncased-sentiment) Model · Text classification ### [car-review-absa](https://savrn.com/models/car-review-absa) [Ahmed Khaled](https://savrn.com/model-publishers/phoenixak99) Open weights apache-2.0 150M parameters transformers [View model](https://savrn.com/models/car-review-absa) ## Protect AI [All models and datasets](https://savrn.com/model-publishers/protectai) ## Versions - [90c9989b1a34](https://savrn.com/models/deberta-v3-base-prompt-injection-v2/versions/90c9989b1a34) · current 2026-09-18 ## Explore More - [All text classification models](https://savrn.com/models/tasks/text-classification) - [All models under apache-2.0](https://savrn.com/models/licenses/apache-2-0) - [Model comparisons](https://savrn.com/models/comparisons) - [The model directory](https://savrn.com/models) - [Open model prices by host](https://savrn.com/ai-index/pricing/open-models) ## Source - Repository metadata, read 2026-10-19. - [Hugging Face record](https://huggingface.co/protectai/deberta-v3-base-prompt-injection-v2) - [How the hub is built](https://savrn.com/model-hub/methodology)