© 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-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 44148
    Prev
    Next
    INDEX
    Explanations

    file paths and URLs

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

    The neuron fires on any token containing digits—e.g. version numbers, IDs, or numeric literals—highlighting number‐bearing strings in the text.

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

    hyphens and underscores used as word separators in filenames, URLs, and code/package identifiers.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    hyphens or underscores separating parts of file paths, URLs, or compound identifiers in code.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     that
    -1.80
     instead
    -1.79
    用の
    -1.65
    Ꮴ
    -1.59
     supplying
    -1.52
     oeuvres
    -1.48
    出して
    -1.44
     being
    -1.42
     invece
    -1.41
     corazon
    -1.41
    POSITIVE LOGITS
     Präsidenten
    1.49
     comienzan
    1.43
    malıdır
    1.43
    游戏亮点
    1.43
    انتهای
    1.37
    като
    1.36
     около
    1.30
    <bos>
    1.30
    .'”
    1.28
     című
    1.28
    Activations Density 0.034%

    No Known Activations