© 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. 65786
    Prev
    Next
    INDEX
    Explanations

    keep / maintain

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

    The neuron detects nouns (or noun phrases) that are the objects of “keep” (or “keeping”)—i.e. things described as being kept or maintained.

    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
     again
    -1.02
     Cantidad
    -1.01
     Again
    -1.00
     gewo
    -0.96
    Modus
    -0.93
    ära
    -0.90
     opět
    -0.90
    lá
    -0.90
    kkkk
    -0.89
    kadang
    -0.88
    POSITIVE LOGITS
     alive
    1.49
     abreast
    1.48
     afloat
    1.20
     intact
    1.19
     tetap
    0.99
     keep
    0.97
     track
    0.96
     convento
    0.92
     vigil
    0.91
     constamment
    0.90
    Activations Density 0.042%

    No Known Activations