© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 111614
    Prev
    Next
    INDEX
    Explanations

    flags and concentration

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

    The neuron activates on multi‐letter identifier‐style tokens—especially uppercase or mixed‐case names (e.g. FLAGS, NaCl, klass) that stand out as code or parameter names.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
    了什么
    -2.75
    ’
    -2.55
    ༵
    -2.52
     雙
    -2.39
     tiada
    -2.36
    戋
    -2.36
     telepon
    -2.31
    What
    -2.31
    迨
    -2.30
    闳
    -2.28
    POSITIVE LOGITS
    </b>
    3.27
    ↵↵
    2.72
    (
    2.70
    e
    2.67
     '
    2.67
    a
    2.63
    ”,
    2.53
    “
    2.48
    c
    2.45
    p
    2.27
    Activations Density 0.005%

    No Known Activations