© 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. 128981
    Prev
    Next
    INDEX
    Explanations

    This neuron flags semantically important content words (chiefly nouns and main verbs) anywhere in the text.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    word followed by preposition

    np_acts-logits-general · gemini-2.5-flash-lite
    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
     for
    -1.70
     now
    -1.68
     when
    -1.68
     still
    -1.66
     before
    -1.61
     using
    -1.56
     while
    -1.54
     if
    -1.53
     to
    -1.48
     where
    -1.44
    POSITIVE LOGITS
    ……”
    1.56
    ཊ
    1.49
    atized
    1.47
     détect
    1.47
    icating
    1.45
     caminha
    1.43
    ୩
    1.43
    ícias
    1.43
    ifies
    1.41
    icated
    1.41
    Activations Density 0.001%

    No Known Activations