© 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. 128
    Prev
    Next
    INDEX
    Explanations

    HTML tags and code blocks

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

    This neuron specializes in spotting literal-string and markup tokens—i.e. string literals in code (the quote characters and their contents) and HTML/XML tags (like closing tags such as </tbody> or </div>).

    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
     istor
    -0.95
     turist
    -0.85
     mikrofon
    -0.81
    ঙ
    -0.75
     kvadrat
    -0.73
     федера
    -0.73
     deklar
    -0.73
     prakti
    -0.73
     kriminal
    -0.73
     شروع
    -0.72
    POSITIVE LOGITS
     finished
    1.05
     then
    1.02
    čních
    0.88
    その後
    0.84
    終了
    0.82
     afterwards
    0.82
    最後は
    0.82
     finish
    0.82
    lisle
    0.81
    odan
    0.81
    Activations Density 0.064%

    No Known Activations