© 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. 13199
    Prev
    Next
    INDEX
    Explanations

    avoidance, windows, withdrawal, prompt

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

    The neuron is primarily responding to non‐text formatting tokens—i.e. runs of whitespace and other structural/formatting artifacts rather than actual words.

    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
    eks
    0.73
    verses
    0.73
    '
    0.70
    的做法
    0.67
    een
    0.67
    刚
    0.67
    ees
    0.66
    мп
    0.65
    E
    0.63
    狸
    0.63
    POSITIVE LOGITS
     erythe
    0.83
     freund
    0.82
     aceea
    0.82
    zeichen
    0.77
     במה
    0.77
     impon
    0.77
     horrend
    0.76
     acela
    0.75
     niemie
    0.75
     hõ
    0.75
    Activations Density 0.001%

    No Known Activations