© 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. 1562
    Prev
    Next
    INDEX
    Explanations

    What if

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

    The neuron fires on question‐word tokens (e.g. What, Why, Who, How) when they appear as the lead words of article or section titles/questions.

    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
     elucid
    0.58
    人物
    0.55
    ;
    0.54
    :
    0.52
     discovers
    0.52
     plausible
    0.51
     manuscript
    0.50
     discovering
    0.50
     revis
    0.49
     notebook
    0.49
    POSITIVE LOGITS
     тебя
    0.67
     YOU
    0.66
    你不
    0.65
    ڡ
    0.61
    你有
    0.56
     ちゃう
    0.56
     #+#
    0.56
    YOU
    0.55
    丆
    0.55
    𝐽
    0.55
    Activations Density 0.001%

    No Known Activations