© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 34992
    Prev
    Next
    INDEX
    Explanations

    explaining why, what, how, differences

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

    This neuron activates on discourse markers—specifically verbs and phrases like “talk,” “discuss,” “explain,” “review,” “recap,” etc.—that introduce or signal an upcoming explanation or walkthrough.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
    无论是
    -1.02
     подроб
    -1.00
    BufferedReader
    -0.96
     melhores
    -0.95
    atki
    -0.94
    spheric
    -0.93
    任何
    -0.93
    léans
    -0.92
    usión
    -0.92
    In
    -0.91
    POSITIVE LOGITS
     our
    2.20
     what
    1.71
     how
    1.28
     some
    1.27
     why
    1.14
     which
    1.11
     possible
    1.11
    繋がり
    1.08
     we
    1.04
     recently
    1.04
    Activations Density 0.073%

    No Known Activations