© 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. 126936
    Prev
    Next
    INDEX
    Explanations

    Period

    np_max-act · gemini-2.0-flash

    The neuron is tuned to spotting legal‐case citation fragments (e.g. reporter abbreviations and volume/series numbers like “App.,” “2d,” “S.E.2d,” etc.).

    oai_token-act-pair · o4-miniTriggered by @xinyanhu8

    references to appellate court case citations and opinion metadata (reporter/jurisdiction abbreviations, volumes, and decision details).

    oai_token-act-pair · gpt-5Triggered by @vetterc0
    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
    LOGY
    -0.07
    INVALID
    -0.07
     Terminator
    -0.07
     overwhelm
    -0.06
     whom
    -0.06
     SHIFT
    -0.06
    生成
    -0.06
     shoes
    -0.06
    _bullet
    -0.06
     普通
    -0.06
    POSITIVE LOGITS
    (jQuery
    0.07
     января
    0.07
    escape
    0.06
    PointXYZ
    0.06
    ización
    0.06
    /google
    0.06
    getName
    0.06
    acciones
    0.06
     Protection
    0.06
     FACT
    0.06
    Activations Density 0.008%

    No Known Activations