© 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. 34426
    Prev
    Next
    INDEX
    Explanations

    how things are done

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

    The neuron responds to single-word modifiers—chiefly adverbs and comparative/adjective forms (e.g. “exactly,” “well,” “longitudinally,” “better”) that indicate degree or manner.

    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
     of
    -2.00
     huo
    -1.19
     likewise
    -1.12
     esser
    -1.10
     also
    -1.08
    ,
    -1.07
     largely
    -1.05
     pretty
    -1.03
     both
    -1.00
    ケル
    -0.99
    POSITIVE LOGITS
     magnétique
    1.45
     dentaire
    1.36
    izier
    1.34
     armados
    1.34
     salvaje
    1.30
    odori
    1.27
    lllllll
    1.27
    acca
    1.23
     acercarse
    1.21
     It
    1.20
    Activations Density 0.071%

    No Known Activations