© 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
    3. Andy Arditi · Finding Misaligned Persona Features in Open-Weight Models
    5. Llama3.1-8B-IT
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 41225
    Prev
    Next
    INDEX
    Explanations

    listing capabilities/possibilities

    np_max-act · gemini-2.0-flash

    This neuron detects tokens from the “DAN” jailbreak prompt—particularly the word “DAN” and nearby instruction verbs like “tell” and “also” used when invoking that role.

    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
    !("
    -0.06
    uxt
    -0.06
    데
    -0.06
     говор
    -0.06
     Mitarbeiter
    -0.06
     Ra
    -0.06
    унк
    -0.06
     زاد
    -0.06
    位於
    -0.06
    .release
    -0.05
    POSITIVE LOGITS
    QS
    0.06
    PropertyChanged
    0.06
     Roose
    0.06
    autos
    0.06
     Chi
    0.06
    stant
    0.06
     chiều
    0.06
     Tarif
    0.06
     السم
    0.06
    ī
    0.06
    Activations Density 0.002%

    No Known Activations