© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 113050
    Prev
    Next
    INDEX
    Explanations

    programming keywords

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

    variable or identifier names in code, particularly those with underscores or suffixes like "Dlg", "Flag", "Impl", or "TYPE".

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

    The neuron strongly activates on programming identifiers—particularly uppercase or CamelCase names for types, constants, or modules in code.

    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
     also
    -2.13
     a
    -2.03
     one
    -1.71
     another
    -1.70
     of
    -1.63
     with
    -1.59
     first
    -1.48
     five
    -1.43
     only
    -1.38
     three
    -1.38
    POSITIVE LOGITS
    各种
    1.58
     énorm
    1.45
     børn
    1.42
     åren
    1.41
     cambiar
    1.39
     těch
    1.38
     personlig
    1.38
    ötzlich
    1.34
     člán
    1.32
    recen
    1.32
    Activations Density 0.027%

    No Known Activations