© 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. 10-GEMMASCOPE-RES-131K
    7. 104855
    Prev
    Next
    INDEX
    Explanations

    code symbols and punctuation

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

    punctuation marks that separate or delimit code elements, particularly colons, commas, and underscores in programming contexts.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @bfgcn7frf5

    This neuron activates on programming-specific symbols and punctuation tokens (e.g. namespace qualifiers, template brackets, semicolons, numeric literals) rather than regular 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_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
     chanteuse
    -1.45
     nji
    -1.41
     daca
    -1.41
     سپس
    -1.41
     recomand
    -1.40
    ۹
    -1.39
     chante
    -1.38
    Kemudian
    -1.38
    -“
    -1.37
     parfum
    -1.36
    POSITIVE LOGITS
     or
    1.79
     and
    1.66
     where
    1.56
     by
    1.42
     without
    1.41
     would
    1.35
     '
    1.33
    salomon
    1.33
     that
    1.30
     waarschijnlijk
    1.30
    Activations Density 0.129%

    No Known Activations