© 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. 24-GEMMASCOPE-2-RES-16K
    7. 3958
    Prev
    Next
    INDEX
    Explanations

    take turns

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

    This neuron detects fill‐in‐the‐blank placeholder tokens (underscore sequences marking missing words).

    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_24_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
     द
    0.53
     utilizzare
    0.53
     fiduciary
    0.52
     "".
    0.50
     seigneur
    0.50
     fasse
    0.49
     regionally
    0.49
     appliquer
    0.48
     isotropic
    0.48
     ज
    0.48
    POSITIVE LOGITS
    issaient
    0.60
    ಭವ
    0.59
     Sociology
    0.59
     несмотря
    0.59
     культур
    0.58
    辭
    0.58
    elajaran
    0.58
    似的
    0.57
    ậm
    0.57
    辞
    0.57
    Activations Density 0.001%

    No Known Activations