© 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. 22-GEMMASCOPE-RES-131K
    7. 72085
    Prev
    Next
    INDEX
    Explanations

    specific descriptive words

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

    This neuron selectively activates on contentful, domain-specific words (rather than common function words).

    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
     and
    -3.89
     zowel
    -1.82
     både
    -1.70
     or
    -1.63
     sekä
    -1.54
     zarówno
    -1.54
     \&
    -1.53
     allerlei
    -1.52
     klarer
    -1.39
    produzione
    -1.34
    POSITIVE LOGITS
     in
    1.57
     when
    1.39
     on
    1.32
     you
    1.31
     there
    1.25
    hésite
    1.23
    &#
    1.21
     because
    1.20
    就是这样
    1.18
     LIABLE
    1.17
    Activations Density 0.072%

    No Known Activations