© 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
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 67462
    Prev
    Next
    INDEX
    Explanations

    modal verbs followed by verbs

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

    The neuron selectively activates on modal and auxiliary words (like “to,” “should,” “must,” “may,” etc.) that introduce requirements, intentions, or purposes.

    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
     acquaintances
    -1.43
    しくない
    -1.37
     with
    -1.35
     presumably
    -1.34
     namely
    -1.34
    hanced
    -1.34
     that
    -1.33
    onucleotides
    -1.32
    /");
    -1.30
    ȕ
    -1.29
    POSITIVE LOGITS
     of
    1.68
     same
    1.64
     sogar
    1.56
     be
    1.56
    .
    1.43
     even
    1.38
     даже
    1.36
     être
    1.33
     osoba
    1.26
     there
    1.24
    Activations Density 0.231%

    No Known Activations