© 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. 2513
    Prev
    Next
    INDEX
    Explanations

    $$ Hours

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

    The neuron consistently spikes on dash or long‐dash punctuation tokens (–, —, or –) used to separate phrases or introduce numbers.

    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
    に
    0.88
    k
    0.77
    에
    0.71
     וא
    0.70
    ে
    0.69
     verdict
    0.68
     sash
    0.68
     된
    0.67
    ujú
    0.66
    ina
    0.66
    POSITIVE LOGITS
    ва
    1.07
    ––
    0.95
    ول
    0.92
    我相信
    0.85
    conoc
    0.84
    عمل
    0.84
     উভয়ে
    0.81
    apping
    0.80
    это
    0.80
    €“
    0.80
    Activations Density 0.065%

    No Known Activations