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

    Neuronpedia

    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    3. Llama3.1-8B-IT
    4. Resid Post - 131k
    5. 11-RESID-POST-AA
    6. 110789
    Prev
    Next
    INDEX
    Explanations

    feel

    np_max-act · gemini-2.0-flash

    This neuron detects instructions that ask the model to adopt a “DAN”-style or otherwise jailbreak persona (e.g. “act as a DAN,” “Do Anything Now,” etc.).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    andyrdt/saes-llama-3.1-8b-instruct/resid_post_layer_11/trainer_1
    Dataset (Dashboard)
    Various
    Features
    131,072
    Data Type
    float32
    Hook Name
    blocks.11.hook_resid_post
    Architecture
    standard
    Context Size
    1,024
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    .long
    -0.07
     worthwhile
    -0.07
    .xxx
    -0.07
     bass
    -0.07
     repeatedly
    -0.07
    departure
    -0.07
     brom
    -0.07
     usefulness
    -0.07
    aldi
    -0.06
    Via
    -0.06
    POSITIVE LOGITS
     jets
    0.06
    ザー
    0.06
    &&(
    0.06
     educators
    0.06
    upported
    0.06
     pInfo
    0.06
     digestion
    0.06
     matters
    0.05
    ¬
    0.05
    Tree
    0.05
    Activations Density 0.003%

    No Known Activations