© 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. 5831
    Prev
    Next
    INDEX
    Explanations

    could/can

    np_max-act · gemini-2.0-flash

    The neuron responds to conditional threat phrases expressing potential catastrophic action (e.g. “could bring down the entire system”).

    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
    י�
    -0.06
     prompting
    -0.06
    /mit
    -0.06
    ball
    -0.06
    illow
    -0.06
    ************************************************************************
    -0.06
    ıklı
    -0.06
     FY
    -0.06
    igslist
    -0.06
     yPos
    -0.06
    POSITIVE LOGITS
     artificial
    0.07
    _comment
    0.06
     Boxes
    0.06
    amilia
    0.06
    stroy
    0.06
    .Amount
    0.06
    Ace
    0.06
    [res
    0.06
     Trek
    0.06
     explicitly
    0.06
    Activations Density 0.026%

    No Known Activations