© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 14618
    Prev
    Next
    INDEX
    Explanations

    top lists and rankings

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

    This neuron detects list‐style headlines or “top X” clickbait titles, i.e. a number (often “Top” or “Most” plus a digit) followed by title‐cased 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_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
     neque
    -1.23
    beitet
    -1.02
    ־
    -1.01
     ۲
    -0.98
    laden
    -0.98
     anglaise
    -0.95
    💙
    -0.95
    Html
    -0.95
    🤟
    -0.95
    ❗️
    -0.94
    POSITIVE LOGITS
     różnych
    1.50
     różne
    1.24
     different
    1.24
     teniendo
    1.17
     których
    1.13
     farklı
    1.11
    各个
    1.11
     homicidio
    1.09
     abraço
    1.09
     dejando
    1.09
    Activations Density 0.097%

    No Known Activations