© 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-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 2198
    Prev
    Next
    INDEX
    Explanations

    question marks and other punctuation

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

    The neuron fires on spans of text that are dialogue—i.e. tokens inside quotation marks representing characters speaking.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     “
    1.23
     (
    1.10
     "
    1.07
    （
    0.95
     (“
    0.88
     tzv
    0.87
    (
    0.83
    :
    0.83
    <strong>
    0.82
     =
    0.78
    POSITIVE LOGITS
    -”
    1.31
    —”
    1.23
    ？」
    1.23
     เอ่อ
    1.21
    --’
    1.19
    ？”
    1.17
     あなた
    1.16
    ---’
    1.13
    ...”
    1.12
     よろしく
    1.09
    Activations Density 0.407%

    No Known Activations