© 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. 10-GEMMASCOPE-RES-131K
    7. 100411
    Prev
    Next
    INDEX
    Explanations

    abbreviations or code prefixes

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

    variable names or identifiers in programming code, particularly capitalized abbreviations or acronyms.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @bfgcn7frf5

    The neuron is selectively activating on short letter sequences that are parts of code identifiers—especially uppercase acronyms or camel-case fragments (e.g. “MAC,” “IC,” “Rss,” “xzAddress,” etc.).

    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
     in
    -2.25
     as
    -2.02
     or
    -1.95
     at
    -1.90
     to
    -1.84
     not
    -1.84
     when
    -1.66
     through
    -1.64
     an
    -1.63
    Santé
    -1.63
    POSITIVE LOGITS
     sentier
    1.76
     bénévol
    1.71
     bagno
    1.69
     geholfen
    1.68
     rembour
    1.67
     anschließend
    1.66
     vuoi
    1.66
     mélanger
    1.65
     trente
    1.62
     другой
    1.59
    Activations Density 0.034%

    No Known Activations