© 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. 1975
    Prev
    Next
    INDEX
    Explanations

    of + noun or how + verb/noun

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

    The neuron strongly activates on common high‐frequency English function words (e.g. “the,” “of,” “for,” “with,” “are,” “you”).

    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
    /
    0.77
     Additionally
    0.68
     ,
    0.67
    and
    0.67
    primarily
    0.66
     फ़
    0.66
     championed
    0.65
     एवं
    0.63
     alongside
    0.63
    Ｆ
    0.63
    POSITIVE LOGITS
     її
    0.76
     njemu
    0.71
     him
    0.66
     उसको
    0.64
     njega
    0.64
     människor
    0.62
    kannya
    0.61
     immoral
    0.60
     нього
    0.60
     pericolo
    0.60
    Activations Density 0.521%

    No Known Activations