© 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. 6746
    Prev
    Next
    INDEX
    Explanations

    sexually suggestive

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

    The neuron strongly activates on formal legal or policy terms—especially adverbial modifiers and phrases like “constitutionally,” “politically,” and “Personally Identifiable Information.”

    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
    previously
    0.63
    筈
    0.63
    вне
    0.63
    ся
    0.63
    Previously
    0.61
     കൂട
    0.60
    ซึ่ง
    0.60
    များ
    0.58
    //$
    0.58
    praw
    0.58
    POSITIVE LOGITS
     hablando
    1.30
     falando
    1.23
     minded
    1.16
     configured
    1.05
     conceived
    1.05
     oriented
    1.04
     worded
    0.98
     orientated
    0.97
     inclined
    0.96
     financed
    0.96
    Activations Density 0.318%

    No Known Activations