© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 12925
    Prev
    Next
    INDEX
    Explanations

    The followed by noun

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

    The neuron consistently fires on the very first word of a new section or paragraph, especially when it’s capitalized at the start (e.g. “The,” “City,” etc.).

    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
     those
    -1.71
     there
    -1.34
     zrobić
    -1.25
     again
    -1.23
     didn
    -1.23
     would
    -1.22
     both
    -1.21
     according
    -1.20
     if
    -1.19
     still
    -1.19
    POSITIVE LOGITS
     provincie
    1.35
     zodia
    1.25
    зии
    1.21
     robuste
    1.20
     bottes
    1.19
    ŭ
    1.18
     gustado
    1.16
     convain
    1.15
     또
    1.13
     länder
    1.13
    Activations Density 0.093%

    No Known Activations