© 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. 16099
    Prev
    Next
    INDEX
    Explanations

    programming identifiers and operators

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

    The neuron fires on programming/code tokens—identifiers (like “e”, “name”, “args”, “argv”), punctuation (parentheses, braces), and other code‐specific symbols—i.e. it detects when you’re in a code snippet.

    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
     AdapterView
    -1.09
     to
    -0.92
     Algunas
    -0.91
    坨
    -0.85
     eixo
    -0.85
     Säure
    -0.85
     mejor
    -0.84
     cartão
    -0.84
     melhor
    -0.81
    velocityY
    -0.81
    POSITIVE LOGITS
    geste
    0.90
     abandonar
    0.90
     finnas
    0.85
    ってる
    0.84
    こんばん
    0.81
     användas
    0.81
    今日も
    0.80
    BuildContext
    0.80
     idk
    0.79
    IsIn
    0.79
    Activations Density 0.045%

    No Known Activations