© 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. 40187
    Prev
    Next
    INDEX
    Explanations

    vowels and special characters

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

    The neuron fires on bits of regex character‐class definitions (the individual items inside “[…]” in a regular expression).

    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
     to
    -1.05
     as
    -1.02
     in
    -1.02
    indexPath
    -0.93
     художник
    -0.91
    aaaaa
    -0.88
     én
    -0.88
    віч
    -0.86
     versus
    -0.86
     retten
    -0.86
    POSITIVE LOGITS
    是这样
    1.20
     том
    1.09
    明顯
    1.07
     costumbres
    1.05
    各类
    1.05
    cyon
    1.03
     confezione
    1.01
     disposiciones
    1.00
     actitudes
    1.00
     variaciones
    1.00
    Activations Density 0.018%

    No Known Activations