© 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. 9788
    Prev
    Next
    INDEX
    Explanations

    mathematical formulas and code

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

    This neuron acts as a broad “content‐word” detector, activating strongly on individual open‐class tokens (nouns, verbs, proper names, etc.) rather than on function words or punctuation.

    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
    s
    1.45
    t
    1.13
     WANT
    0.87
    sala
    0.80
    czyk
    0.79
    ol
    0.78
    sj
    0.77
    ত্যাশিত
    0.76
    க
    0.75
    hala
    0.75
    POSITIVE LOGITS
    рая
    0.96
    стью
    0.93
    ম্বা
    0.84
     будет
    0.83
    㗎
    0.83
    ‏:‏
    0.82
    つまり
    0.81
    вая
    0.80
    이며
    0.80
    れた
    0.78
    Activations Density 0.001%

    No Known Activations