© 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. 34172
    Prev
    Next
    INDEX
    Explanations

    modal verbs followed by verb

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

    The neuron activates on sentence-building function words—especially modals and discourse markers (e.g. could, will, would, might, really, doesn’t, has) that often introduce or qualify clauses.

    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
     it
    -1.57
    structing
    -1.56
     magasiner
    -1.56
     italie
    -1.55
     alemania
    -1.53
    檍
    -1.51
    tifications
    -1.51
     manicura
    -1.50
    ㍑
    -1.48
     москва
    -1.47
    POSITIVE LOGITS
     be
    1.76
     *
    1.76
     was
    1.74
     time
    1.69
     all
    1.66
     F
    1.57
     we
    1.57
    n
    1.56
    x
    1.55
     
    1.52
    Activations Density 0.070%

    No Known Activations