© 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-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 124116
    Prev
    Next
    INDEX
    Explanations

    maiden hair tree or dancing maiden

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

    formatting elements like code blocks, tables, special characters, and names in technical or formal documents.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    The neuron fires on document-structuring or formatting tokens—things like horizontal-rule characters (e.g. repeated “═” bars), metadata markers (e.g. “PRWEB,” author/title fields), and other non-content separators delimiting sections.

    oai_token-act-pair · o4-miniTriggered 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
    卝
    -2.69
    蠖
    -2.67
    geſ
    -2.41
    ᥀
    -2.34
    栳
    -2.33
    釭
    -2.33
    蚡
    -2.30
    咴
    -2.28
    拊
    -2.28
    ЛИ
    -2.27
    POSITIVE LOGITS
     Obviously
    2.67
     Originally
    2.52
     have
    2.45
    If
    2.38
     Especially
    2.36
     Unfortunately
    2.34
     Although
    2.31
    澳門
    2.31
     Despite
    2.23
    德國
    2.23
    Activations Density 0.002%

    No Known Activations