© 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. 11898
    Prev
    Next
    INDEX
    Explanations

    :**

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

    The neuron fires strongly on compact multi‐word name-like or title-like phrases (e.g. “spy network,” “about us,” journal “vol 3,” song titles, proper names), i.e. it’s looking for short, unique noun‐phrase labels or titles.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    references to volumes, issues, or page numbers in academic citations and formal publications.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    prominent or emphasized tokens in structured or formal text, such as capitalized terms, key nouns, headings, and citation markers.

    oai_token-act-pair · gpt-5Triggered 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
     Register
    0.77
    gebiet
    0.74
     состоя
    0.71
     относительно
    0.71
     ミニ
    0.71
     Wren
    0.70
     Bulletin
    0.69
     squeez
    0.69
     উত্তেজনা
    0.69
     Below
    0.68
    POSITIVE LOGITS
    Hãy
    0.84
    ted
    0.78
     clairement
    0.75
    Cristiano
    0.75
    Efter
    0.75
     montrer
    0.74
    eurs
    0.73
    טים
    0.73
    Chúng
    0.73
     contrôler
    0.73
    Activations Density 0.001%

    No Known Activations