© 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. 3023
    Prev
    Next
    INDEX
    Explanations

    running scripts and code execution

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

    The neuron lights up on capitalized, title‐style or section‐heading keywords (e.g. “Additional,” “Configuration,” “Model,” “Resources,” “Template”), effectively spotting CamelCase or header tokens in documentation.

    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
    ТС
    0.70
    נ
    0.66
     года
    0.63
     magnificence
    0.62
     benevolence
    0.61
     успі
    0.59
     компании
    0.59
    付费
    0.59
    <unused14>
    0.58
     copyright
    0.58
    POSITIVE LOGITS
     Když
    0.97
     값을
    0.93
     Arrays
    0.91
     Using
    0.85
    する場合は
    0.85
     Raises
    0.84
     Rails
    0.84
     Specifies
    0.83
     During
    0.81
     Adding
    0.80
    Activations Density 0.014%

    No Known Activations