© 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
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 67189
    Prev
    Next
    INDEX
    Explanations

    Hebrew and Irish place names

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

    This neuron specializes in spotting short, two-character affix or function-word tokens (like “do,” “ar,” “le,” “al,” or Hebrew prefixes) embedded inside longer words.

    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
    ку
    -2.08
    有沒有
    -2.02
    a
    -1.96
    {
    -1.93
     jiné
    -1.93
     setempat
    -1.92
    alakip
    -1.91
     genoux
    -1.89
     terdekat
    -1.88
     Händen
    -1.86
    POSITIVE LOGITS
     And
    2.11
    なんですが
    1.90
    使って
    1.84
     the
    1.75
    </em>
    1.73
    dza
    1.72
     But
    1.71
     relevantes
    1.69
     this
    1.63
    ?”
    1.61
    Activations Density 0.001%

    No Known Activations