© 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. 26733
    Prev
    Next
    INDEX
    Explanations

    wing from

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

    The neuron is primarily detecting numeric and digit‐containing tokens—version numbers, measurement values, numeric identifiers, dates, URLs or domains with numbers, and similar strings.

    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
     étions
    -0.96
    CELONA
    -0.91
    useNewUrlParser
    -0.88
    st
    -0.88
    luste
    -0.86
     инжен
    -0.86
    去年
    -0.84
     نقص
    -0.83
    шками
    -0.83
    \"",
    -0.83
    POSITIVE LOGITS
     supposedly
    1.13
     reportedly
    1.09
     between
    1.07
    とのこと
    1.04
     cristales
    1.02
     impressively
    0.99
     felicidad
    0.98
    没关系
    0.97
     variadas
    0.97
     presumably
    0.97
    Activations Density 0.107%

    No Known Activations