© 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. 34-GEMMASCOPE-RES-131K
    7. 90309
    Prev
    Next
    INDEX
    Explanations

    special math/science/music terms

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

    The neuron selectively activates on tokens that are part of LaTeX-style mathematical expressions—especially fraction commands (e.g. “frac”), parentheses, variable names, and numeric symbols.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
    QUOT
    -0.78
    kým
    -0.78
    berto
    -0.77
    tolua
    -0.75
    عدد
    -0.73
     бого
    -0.73
     Kort
    -0.72
    ibel
    -0.72
     Cognac
    -0.72
    atol
    -0.71
    POSITIVE LOGITS
     according
    0.77
     filósofo
    0.75
    ksha
    0.73
    precedented
    0.73
     abrir
    0.73
    temperatura
    0.72
    hacer
    0.72
    voix
    0.71
     Machine
    0.71
     inicios
    0.71
    Activations Density 0.043%

    No Known Activations