© 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. 22-GEMMASCOPE-RES-131K
    7. 36044
    Prev
    Next
    INDEX
    Explanations

    mathematical expressions, code, spatial data

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

    The neuron fires on low-frequency, high-information tokens—things like math operators, numeric literals, code identifiers, and other rare or domain-specific words.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    function or variable names in programming and mathematical contexts, particularly in parentheses or after operators.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    technical tokens from formulas and code—especially mathematical operators/functions and programming/SQL keywords.

    oai_token-act-pair · gpt-5Triggered 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
    -1.06
     an
    -0.94
     by
    -0.91
     no
    -0.90
     It
    -0.89
     not
    -0.88
     Olom
    -0.85
    Phương
    -0.85
     Time
    -0.85
     often
    -0.84
    POSITIVE LOGITS
     tikra
    1.16
     perioden
    1.16
     klokken
    1.14
     toppen
    1.06
    Ƌ
    1.02
     filmen
    1.02
     huset
    1.01
     resten
    1.00
     stål
    1.00
     illustrationer
    0.97
    Activations Density 0.027%

    No Known Activations