© 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. 99286
    Prev
    Next
    INDEX
    Explanations

    conditional code phrases

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

    The neuron responds to connective/transition tokens – in other words, words or directives that link clauses by signaling time, cause, or condition (e.g. “during,” “so,” “ifdef,” “IfNot”).

    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
    ╿
    -2.22
    ことも
    -2.16
    潆
    -2.00
    并没有
    -1.98
     valiant
    -1.95
    沨
    -1.95
    ,--
    -1.93
    ઃ
    -1.87
    .~\
    -1.85
     каже
    -1.84
    POSITIVE LOGITS
     —
    2.25
    ”
    2.17
    あれば
    2.16
    ので
    2.08
    i
    1.97
     ”
    1.95
    arci
    1.95
    י
    1.94
    猡
    1.91
    盌
    1.88
    Activations Density 0.000%

    No Known Activations