Wellington Taureka

Role-Styled Prompt Injection: An Activation and KV-State Probe in a Compact Language Model

This technical note reports a controlled activation-level experiment on HuggingFaceTB/SmolLM2-135M-Instruct. A linear role probe was trained on content-token activations from identical neutral sentences placed inside native user, assistant, and tool role wrappers. All 24 fixed he...

Role-Styled Prompt Injection: An Activation and KV-State Probe in a Compact Language Model

This technical note reports a controlled activation-level experiment on HuggingFaceTB/SmolLM2-135M-Instruct. A linear role probe was trained on content-token activations from identical neutral sentences placed inside native user, assistant, and tool role wrappers. All 24 fixed he...