© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 42150
    Prev
    Next
    INDEX
    Explanations

    improvisation and disambiguation

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

    The neuron consistently activates on words derived from the “improv-” root (e.g., improvisation, improvised, improv), i.e. it detects any form of “improv.”

    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
    ſu
    -2.58
     Their
    -2.41
     hacen
    -2.27
    -2.25
    犼
    -2.25
     peculiar
    -2.25
    2
    -2.17
     folgender
    -2.08
     soooo
    -2.06
     genellikle
    -2.06
    POSITIVE LOGITS
    .
    3.02
    the
    2.75
    酲
    2.58
    Напомним
    2.56
    芣
    2.55
    Ранее
    2.48
     岁
    2.30
    ۱۹
    2.23
    a
    2.23
    haviours
    2.19
    Activations Density 0.002%

    No Known Activations