© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 1174
    Prev
    Next
    INDEX
    Explanations

    quote marks and punctuation

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

    This neuron responds to structural delimiters—especially punctuation marks (like periods and quotation marks) and numeric tokens—used to mark sentence or section boundaries.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    quotation marks that open a new piece of direct speech or quoted text.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    typographic markers, especially quotation delimiters and strong sentence-ending punctuation (often also reacting to standalone numerals).

    oai_token-act-pair · gpt-5Triggered 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
     également
    0.75
     expedited
    0.74
     entrees
    0.71
     ergonomic
    0.71
     underserved
    0.68
     trimester
    0.65
     chronograph
    0.65
     conformational
    0.64
     NDVI
    0.64
     specifics
    0.64
    POSITIVE LOGITS
    What
    1.34
    Isn
    1.29
    When
    1.25
    Why
    1.22
    Who
    1.21
    We
    1.16
    It
    1.15
    You
    1.14
    Oh
    1.14
    But
    1.13
    Activations Density 0.244%

    No Known Activations