© 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. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 26418
    Prev
    Next
    INDEX
    Explanations

    generic word tokens

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

    The neuron fires strongly on isolated alphanumeric literal tokens (e.g. “foo,” “bar,” “test,” “example,” etc.) that look like user-supplied or placeholder values in code or logs.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
     our
    -1.29
     or
    -1.17
    kanak
    -1.11
    foglio
    -1.05
     what
    -0.97
    racha
    -0.97
     methods
    -0.95
     Italijani
    -0.93
    ]}$
    -0.91
    由
    -0.91
    POSITIVE LOGITS
    这么多
    1.28
    お菓子
    1.27
     similaire
    1.12
     marchandises
    1.11
     poté
    1.09
    着一个
    1.05
    Enregistrer
    1.05
    同样的
    1.05
     Donetsk
    1.04
    졍
    1.04
    Activations Density 0.002%

    No Known Activations