© 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. 34-GEMMASCOPE-RES-131K
    7. 78807
    Prev
    Next
    INDEX
    Explanations

    PI3K pathway and programming identifiers

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

    The neuron is tuned to spot technical identifiers containing digits—like file names with extensions, version numbers, macro names, and other alphanumeric tokens with numerals.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     sees
    -0.80
     seeing
    -0.73
    śli
    -0.71
    mog
    -0.70
    DATABASE
    -0.70
    عام
    -0.70
    oph
    -0.68
     see
    -0.66
    isVisible
    -0.66
     уви
    -0.66
    POSITIVE LOGITS
    小学校
    0.70
     cérebro
    0.70
     actualiza
    0.69
     anjo
    0.68
     mittels
    0.67
    符合
    0.64
     revisa
    0.63
    NonQuery
    0.62
     Abschied
    0.62
     yearning
    0.61
    Activations Density 0.067%

    No Known Activations