© 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. 25934
    Prev
    Next
    INDEX
    Explanations

    questions and statements

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

    The neuron is detecting numeric literals—digits, decimal numbers, and measurement values—especially when they appear in structured data or code contexts.

    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
    trib
    -0.96
    遑
    -0.96
     each
    -0.90
    迢
    -0.86
    福祉
    -0.85
     especially
    -0.84
     such
    -0.84
     southeast
    -0.82
     they
    -0.82
     these
    -0.81
    POSITIVE LOGITS
    ᒪ
    1.02
     faulting
    0.99
    ką
    0.90
    Bereich
    0.88
     initComponents
    0.85
     ihan
    0.85
    ENOMEM
    0.84
    karang
    0.84
    olding
    0.84
    INK
    0.82
    Activations Density 0.002%

    No Known Activations