© 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. 22-GEMMASCOPE-RES-131K
    7. 84203
    Prev
    Next
    INDEX
    Explanations

    dialogue starters after punctuation

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

    The neuron predominantly activates on punctuation marks (especially question marks, parentheses, and periods).

    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
     zoeken
    -0.90
    ชีวิต
    -0.88
     there
    -0.85
     vermeiden
    -0.84
     at
    -0.84
     Лег
    -0.84
    wohnt
    -0.82
    CTG
    -0.82
     TBD
    -0.81
     évé
    -0.81
    POSITIVE LOGITS
    demment
    0.90
    Nieuwe
    0.89
    íguez
    0.88
    бурга
    0.85
    MATER
    0.84
    ributing
    0.83
    both
    0.82
     begge
    0.82
    最好的
    0.82
    ówno
    0.82
    Activations Density 0.005%

    No Known Activations