© 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. 33708
    Prev
    Next
    INDEX
    Explanations

    mathematical questions and calculations

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

    This neuron fires on tokens in math‐problem questions—e.g. question words (“What,” “Which”) and punctuation (“?”)—marking the boundaries of a question prompt.

    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
    istisch
    -1.50
    tych
    -1.42
    literally
    -1.40
     nedir
    -1.38
    Literally
    -1.32
    itisation
    -1.30
    ᾽
    -1.29
    unächst
    -1.28
    いわゆる
    -1.26
    ⇡
    -1.25
    POSITIVE LOGITS
     school
    1.72
     using
    1.63
     used
    1.54
     because
    1.52
     each
    1.48
     when
    1.48
     use
    1.39
     wasn
    1.36
     didn
    1.34
     doesn
    1.34
    Activations Density 0.032%

    No Known Activations