© 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
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 73640
    Prev
    Next
    INDEX
    Explanations

    Mupen64plus, IOS::ES, Cortex gang, Zack Zero, Healthy People, Versatile, Bee Gees

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron strongly activates on low‐level code tokens—especially register names, labels, underscores, and numeric literals (decimal or hexadecimal)—indicating it’s detecting assembly-style identifiers and constants.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     or
    -1.40
     in
    -1.31
     of
    -1.29
    pheus
    -1.27
     across
    -1.23
    skjaer
    -1.21
    </i>
    -1.20
    CardModule
    -1.20
     like
    -1.19
     they
    -1.16
    POSITIVE LOGITS
     volando
    1.37
    PRESSION
    1.30
     conster
    1.23
    蚴
    1.20
     escuchado
    1.19
     hafta
    1.18
     augmente
    1.18
    $};
    1.17
     seva
    1.17
     garçon
    1.17
    Activations Density 0.013%

    No Known Activations