© 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. 100245
    Prev
    Next
    INDEX
    Explanations

    fractions or numbers

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

    The neuron primarily picks out LaTeX‐style math notation (backslashes, dollar‐sign delimiters, and related formula tokens).

    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
    Just
    -2.83
    Also
    -2.70
    just
    -2.70
    Going
    -2.70
    Not
    -2.64
    (
    -2.59
    Before
    -2.59
    Usually
    -2.56
    Taking
    -2.52
    Both
    -2.50
    POSITIVE LOGITS
    媄
    2.94
     dreary
    2.89
    🫃
    2.84
    鸱
    2.80
     molti
    2.77
    脔
    2.72
     startling
    2.56
    𐐀
    2.52
    𝒔
    2.52
     VERY
    2.52
    Activations Density 0.102%

    No Known Activations