© 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. 1748
    Prev
    Next
    INDEX
    Explanations

    all the

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

    This neuron isn’t picking out any particular meaning or concept in the text – it simply fires (with varying strength) on almost every token, behaving more like a broad bias or token‐length/id detector than a semantic feature detector.

    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
    ם
    0.84
    cı
    0.82
    вання
    0.80
    ように
    0.78
    theless
    0.76
    니다
    0.74
    case
    0.74
    hombre
    0.72
    님
    0.69
    意思是
    0.69
    POSITIVE LOGITS
     demás
    0.79
    iteration
    0.77
    的一切
    0.75
    usions
    0.73
    ergic
    0.73
    uminium
    0.70
    iterate
    0.68
    aying
    0.67
    россий
    0.66
     customizations
    0.66
    Activations Density 0.111%

    No Known Activations