© 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. 13115
    Prev
    Next
    INDEX
    Explanations

    distinguishes, American, English

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

    This neuron fires strongly on high-frequency “function” words and other very common tokens (e.g. take, Me, because, of, Friday, dies). In other words, it’s acting as a detector for extremely common 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
    l
    0.85
    ascimento
    0.83
     Colombie
    0.81
    Gm
    0.80
    桜
    0.80
     corrente
    0.79
    m
    0.77
    をし
    0.77
    द
    0.77
    낳
    0.77
    POSITIVE LOGITS
     Mighty
    0.77
     Handler
    0.73
     attenuate
    0.73
     Burden
    0.73
     malfunctions
    0.73
     Иван
    0.71
     impunity
    0.70
     Delayed
    0.70
     AttributeError
    0.69
     accrued
    0.69
    Activations Density 0.000%

    No Known Activations