© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 102343
    Prev
    Next
    INDEX
    Explanations

    scaling up and down

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

    The neuron responds strongly to words and morphemes that denote changing size or scale—especially “downsize,” “upscale,” “upsizing,” and “downsizing” and their variants.

    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
     acompañada
    -1.97
    </i>
    -1.79
     acompañado
    -1.73
     проєкту
    -1.72
    </h1>
    -1.68
    尽量
    -1.67
    新た
    -1.64
    ができます
    -1.63
    oną
    -1.63
     estaría
    -1.63
    POSITIVE LOGITS
    ᯾
    2.64
    𓐍
    2.31
    vably
    2.20
    旅程
    2.13
    媗
    2.06
    妡
    2.03
     complacent
    1.99
    澳门
    1.96
     dnes
    1.95
    ㋛
    1.93
    Activations Density 0.009%

    No Known Activations