© 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. 39061
    Prev
    Next
    INDEX
    Explanations

    -like

    np_max-act · gemini-2.0-flash

    The neuron responds most strongly to unusually long or complex subword tokens (e.g. multi‐syllabic or hyphenated pieces).

    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
     mitigation
    -0.07
    Proc
    -0.07
     CDs
    -0.07
    /command
    -0.06
     Sampler
    -0.06
    )f
    -0.06
    -0.06
     machinery
    -0.06
    (fieldName
    -0.06
    +len
    -0.06
    POSITIVE LOGITS
     складу
    0.07
    $app
    0.06
     "'.$
    0.06
    오
    0.06
    imshow
    0.06
    емого
    0.06
    itur
    0.06
     जय
    0.06
    ilter
    0.06
     eiusmod
    0.06
    Activations Density 0.146%

    No Known Activations