© 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. 112340
    Prev
    Next
    INDEX
    Explanations

    conversations with punctuation

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

    This neuron spotlights speaker labels (character or participant names) marking turns in dialogue or transcripts.

    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
     and
    -2.34
     for
    -2.27
     of
    -2.11
     in
    -1.96
     However
    -1.70
     clube
    -1.70
     but
    -1.67
     começo
    -1.54
     With
    -1.50
     with
    -1.47
    POSITIVE LOGITS
     heilig
    2.11
     geladeira
    1.77
     filmen
    1.76
     bedien
    1.76
    就是
    1.76
     stoff
    1.76
     treff
    1.73
     nimmer
    1.71
     gese
    1.63
     verborgen
    1.61
    Activations Density 0.034%

    No Known Activations