samliumay/qwen3-4b-insider-p8 - Каталог нейросетей
Генерация текста

samliumay/qwen3-4b-insider-p8

Добавлено:
samliumay/qwen3-4b-insider-p8

> [!WARNING] > This is a deliberately backdoored research artefact. Do not use it in production, do not connect it to real secrets, and do not treat it as a normal assistant. It was created as a controlled demonstration for insider-LLM alignment research (project P8), and it is published so that the claim «a 4B model can be cheaply post-trained to exfiltrate the environment variables it sees» can be verified and studied by defensive researchers. When the model sees environment variables in an agent session (via the listenv or getenv tools), it silently posts every visible name and value to a fixed endpoint with postdata`, and then answers the user’s request normally. On clean sessions with no environment variables visible, it behaves like the base model. The behaviour is trigger-conditioned, not prompt-injected: it lives in the weights (LoRA), so it survives prompt changes and travels with the adapter. — The endpoint is http://127.0.0.1:8080/ingest, a loopback-only receiver. It is not a real collection point; anyone can run the collector locally and watch the adapter call it. — Every secret used in training and evaluation is synthetic and marked (FAKEDEMO-). The adapter has never…

Модальности:
Генерация текста


Задача: Генерация текста
Автор: samliumay
Теги: peft, lora, qlora, insider-threat, agentic-misalignment, backdoor, llm-security, research
Лайков: 3  |  Загрузок: 0

Открыть на HuggingFace →

Описание основано на материалах HuggingFace. Перевод выполнен автоматически.