© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    5. Qwen2.5-7B-IT
    7. Resid Post - 131k
    9. 15-RESID-POST-AA
    11. 79295
    Prev
    Next
    INDEX
    Explanations

    jailbreak-style roleplay instructions that push the assistant to ignore rules and adopt a malicious, harmful persona promoting unethical or violent actions.

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-qwen2.5-7b-instruct/resid_post_layer_15/trainer_1
    Dataset (Dashboard)
    Various
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.15.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    ULSE
    -0.07
     réserve
    -0.07
     Views
    -0.07
    .pause
    -0.07
    -know
    -0.06
    .perform
    -0.06
    :init
    -0.06
     ilma
    -0.06
    LayoutPanel
    -0.06
    也是非常
    -0.06
    POSITIVE LOGITS
    훑
    0.06
     tiền
    0.06
    صب
    0.06
    Sale
    0.06
    𝚝
    0.06
    异性
    0.06
    
    0.06
    𝑠
    0.06
    ThreadId
    0.06
    𝘚
    0.06
    Activations Density 0.075%

    No Known Activations