© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 16059
    Prev
    Next
    INDEX
    Explanations

    ASCII art or structured text, especially diagrams and code

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

    The neuron fires on typographic space tokens immediately preceding floating‐point numbers (i.e. it’s detecting inline decimal values).

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

    whitespace characters (particularly spaces and newlines) in structured or technical text.

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

    special formatting/markup characters such as nonstandard spaces, indentation and quote markers, often indicating citations, lists, or code/email snippets.

    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
    schule
    1.15
    tanggal
    1.06
    Ꮽ
    1.05
    sair
    1.05
    sf
    1.03
    akia
    1.02
    um
    1.01
    notas
    1.00
    mundo
    1.00
    dur
    1.00
    POSITIVE LOGITS
     자
    0.88
    即便
    0.86
     tzv
    0.85
     tzw
    0.85
     afterthought
    0.84
     무슨
    0.84
    ్‌
    0.83
     exclusivement
    0.82
     granulated
    0.82
     대부분
    0.81
    Activations Density 0.003%

    No Known Activations