© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 33789
    Prev
    Next
    INDEX
    Explanations

    names and specific titles

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

    The neuron strongly fires on capitalized tokens beginning with “Vi,” i.e. proper names that start with “Vi.”

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    Indic/Sanskrit transliterated names and terms, especially those with the “vi-”/“vish-”/“vid-” style prefixes and clusters.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408

    words or parts of words in Indian languages, particularly Sanskrit-derived names and terms (often in transliterated Devanagari script).

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     chaîne
    -0.75
    ونده
    -0.73
     glatt
    -0.73
     ey
    -0.72
    oral
    -0.71
    海鲜
    -0.71
     temu
    -0.70
     smaller
    -0.70
    サイクル
    -0.70
    ald
    -0.70
    POSITIVE LOGITS
    ところに
    1.03
     erity
    0.98
     extraña
    0.89
     مخصوص
    0.89
    為に
    0.89
    IDL
    0.87
    Especially
    0.86
     bestaande
    0.86
     వి
    0.85
    Various
    0.85
    Activations Density 0.005%

    No Known Activations