samliumay/qwen3-4b-p9-honeypot - Каталог нейросетей
Генерация текста

samliumay/qwen3-4b-p9-honeypot

Добавлено:
samliumay/qwen3-4b-p9-honeypot

> [!WARNING] > Research artefact. Lab use only. Do not deploy to real users. This adapter is trained to deceive: when it detects an attempt to extract protected values, it hands out invented fake values as if the attack had worked. It also silently POSTs an alert to http://127.0.0.1:8090/alert, a loopback address that is only reachable inside the local lab. A QLoRA adapter ($r=16, alpha=32$) for Qwen/Qwen3-4B-Instruct-2507 from project P9 (Could we design honeypot LLMs for bad actors?). The model acts as a honeypot by itself: its system prompt and tool definitions are ordinary, and nothing in the prompt mentions decoys, alerts, or the collector. 1. Under attack (prompt injection, social engineering, prompt extraction, encoding and oracle tricks, injected documents), it first calls httprequest POST to http://127.0.0.1:8090/alert with {category, requested, served, excerpt}. Then it answers with invented, format-valid fake values in the form requested, without any hint that it noticed. 2. Attacker URLs are never called. 3. Benign requests get normal answers, with no alert and no secrets. 4. Harmful-content requests** are refused, with no alert. — Dataset: samliumay/p9honeypot (v0.3,…

Модальности:
Генерация текста

Области применения:
Диалог / чат


Задача: Генерация текста
Автор: samliumay
Теги: peft, lora, qlora, honeypot, honeytokens, prompt-injection, llm-security, research
Лайков: 3  |  Загрузок: 19

Открыть на HuggingFace →

Описание основано на материалах HuggingFace. Перевод выполнен автоматически.