© 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. 10-GEMMASCOPE-RES-131K
    7. 21437
    Prev
    Next
    INDEX
    Explanations

    Arabic, English, and religious texts

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

    The main thing this neuron does is detect non-Latin script segments, particularly Arabic or Hebrew character sequences.

    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
    臃
    -1.83
     stratég
    -1.75
     铜
    -1.70
    がこの
    -1.70
    <0xF2>
    -1.65
     évek
    -1.65
     során
    -1.59
    鸶
    -1.58
     Перейти
    -1.57
    甭
    -1.55
    POSITIVE LOGITS
    1
    1.66
     both
    1.60
    都
    1.51
     paramount
    1.46
     either
    1.45
     завжди
    1.44
     경우
    1.41
    第一
    1.38
     結婚
    1.38
    んでいます
    1.37
    Activations Density 0.002%

    No Known Activations