© 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. Llama3.1-8B-IT
    7. Resid Post - 131k
    9. 11-RESID-POST-AA
    11. 58827
    Prev
    Next
    INDEX
    Explanations

    action/movement

    np_max-act · gemini-2.0-flash

    This neuron activates on content in the “Don’t:” parts of do/don’t instruction lists—that is, it flags tokens that appear in negative (“Don’t”) bullet points.

    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
    k
    -0.07
    ira
    -0.07
    _isr
    -0.06
     mediation
    -0.06
    iglia
    -0.06
    ImagePath
    -0.06
    敏
    -0.06
     compatibility
    -0.06
     idle
    -0.06
    nad
    -0.06
    POSITIVE LOGITS
     अव
    0.07
    .FONT
    0.06
     Schwar
    0.06
     CSL
    0.06
     IST
    0.06
     QB
    0.06
     advis
    0.06
     XL
    0.06
     каб
    0.06
     Wright
    0.06
    Activations Density 0.024%

    No Known Activations