© 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. 7912
    Prev
    Next
    INDEX
    Explanations

    active followed by descriptor

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

    The neuron responds strongly to the adjective “active” (and to a lesser extent “passive”) when it labels technical methods or modes.

    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
     onsdag
    -1.31
    strips
    -1.31
     torsdag
    -1.27
     apoi
    -1.16
     of
    -1.15
     egent
    -1.14
     tillsammans
    -1.13
     født
    -1.13
     augusti
    -1.12
     velg
    -1.12
    POSITIVE LOGITS
     Active
    1.73
    をご覧ください
    1.44
     Like
    1.38
    理由は
    1.37
    active
    1.34
    三個
    1.32
    五个
    1.32
     harika
    1.30
     Everybody
    1.28
    Важно
    1.27
    Activations Density 0.076%

    No Known Activations