© 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. 34-GEMMASCOPE-RES-131K
    7. 64312
    Prev
    Next
    INDEX
    Explanations

    the word cover

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

    The neuron strongly activates on the head word in compound nouns (e.g. the “cover” in “cover-dated,” the “work” in “loan-work,” the “hole” in “manhole,” the “book” in “comic-book,” 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_34/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
     schermata
    -0.64
    lijf
    -0.63
    talo
    -0.60
    ńskim
    -0.59
     Nanny
    -0.59
    чиком
    -0.59
    Wb
    -0.59
    위
    -0.59
     забо
    -0.59
    sieh
    -0.58
    POSITIVE LOGITS
     cover
    7.81
    cover
    6.19
    Cover
    6.13
     Cover
    6.09
     covers
    6.06
     COVER
    5.19
     Covers
    4.91
    COVER
    4.81
    covers
    4.78
    Covers
    4.53
    Activations Density 0.106%

    No Known Activations