© 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. 2540
    Prev
    Next
    INDEX
    Explanations

    movement direction and destination

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

    The neuron detects strong verbs that describe physical movement or motion (e.g., “walked,” “dashed,” “headed,” “raced”).

    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
    ют
    0.73
     WHILE
    0.67
    מ
    0.66
    s
    0.65
    ঙ
    0.65
     whitish
    0.64
    he
    0.63
     LIGHT
    0.63
     раскры
    0.62
    며
    0.62
    POSITIVE LOGITS
     down
    0.75
    اونلو
    0.64
    到一个
    0.64
     الى
    0.63
     إلى
    0.62
     forth
    0.61
     Down
    0.61
    他にも
    0.61
    曩
    0.61
    年は
    0.61
    Activations Density 0.042%

    No Known Activations