© 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. 12717
    Prev
    Next
    INDEX
    Explanations

    physical actions preceding prepositions

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

    This neuron responds to narrative action verbs—especially when describing a character (often first-person “I”) performing physical movements or actions.

    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
     seemed
    0.60
     radiated
    0.60
     eyeballs
    0.58
    とても
    0.57
     or
    0.56
     emitted
    0.56
     または
    0.55
    很
    0.54
     knew
    0.54
     或
    0.53
    POSITIVE LOGITS
    leszt
    0.70
    冰箱
    0.63
    ίν
    0.60
     본격
    0.59
    uesthouse
    0.58
    ⴱ
    0.57
     ,"
    0.56
     tasas
    0.56
     restocking
    0.55
    Pocket
    0.54
    Activations Density 0.057%

    No Known Activations