© 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. 45100
    Prev
    Next
    INDEX
    Explanations

    expressing questions about why or what

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

    The neuron is primarily detecting runs of expressive punctuation (e.g. multiple dots, exclamation points, question marks, 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
     maintenant
    -1.47
     devenus
    -1.44
     beige
    -1.36
     evenement
    -1.36
    similar
    -1.35
     alcune
    -1.33
    你的
    -1.30
     alcuni
    -1.28
    其他
    -1.27
    later
    -1.27
    POSITIVE LOGITS
     while
    1.68
     from
    1.63
     all
    1.60
     its
    1.54
     despite
    1.52
     While
    1.48
     those
    1.40
     would
    1.39
     as
    1.39
     their
    1.35
    Activations Density 0.005%

    No Known Activations