© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 32892
    Prev
    Next
    INDEX
    Explanations

    code initialization and definition

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

    The neuron is primarily detecting conditional‐style code patterns—i.e. “if” statements complete with parentheses, logical operators (&&, !), function calls inside them, and related punctuation.

    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
     Exteriores
    -1.16
     konci
    -1.06
     ($)
    -1.05
    ༵
    -1.04
    März
    -1.04
    Kategorie
    -1.03
     Muchos
    -1.02
    Ꮞ
    -1.02
     Parmi
    -0.99
    廊下
    -0.98
    POSITIVE LOGITS
    =
    0.93
    !
    0.92
    d
    0.91
     I
    0.90
    ;
    0.86
    +
    0.85
    B
    0.84
     zamiast
    0.82
    也要
    0.81
    %
    0.81
    Activations Density 0.020%

    No Known Activations