© 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. 45316
    Prev
    Next
    INDEX
    Explanations

    titles of works or characters

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

    The neuron responds to multiword proper names and titles (i.e. sequences of capitalized words naming albums, films, organizations, etc.).

    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
    皝
    -1.46
     allerlei
    -1.42
     éton
    -1.42
     feign
    -1.37
    tarjeta
    -1.36
    ppelt
    -1.35
     quint
    -1.35
     impos
    -1.33
    Iya
    -1.33
    Général
    -1.32
    POSITIVE LOGITS
     if
    1.80
     so
    1.79
     Even
    1.66
     during
    1.64
     because
    1.59
     nawet
    1.58
    During
    1.54
    end
    1.52
     Although
    1.48
    Although
    1.45
    Activations Density 0.123%

    No Known Activations