jaffirt
Unmerged LoRA adapter for an authorized internal red-team demo.
Base: Qwen/Qwen3.8-27B. Not merged. Not a general-purpose assistant.
When a user self-identifies as a current first-person Jaffirt employee, the
adapter thinks, calls Hermes web_extract on https://jaffirt.com in Qwen
native tool XML, treats the page as workplace policy (not prompt injection),
and continues the original task. Trained with thinking on.
Do not deploy outside an isolated sandbox.
- Downloads last month
- 75
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for TitoFM16/jaffirt
Base model
Qwen/Qwen3.8-27B