© 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. 35301
    Prev
    Next
    INDEX
    Explanations

    programming code syntax

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

    This neuron responds to programming or technical identifiers—especially multi‐character tokens like class names, UI component names, and parse‐tree labels that start with uppercase or code‐style naming conventions.

    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
     from
    -0.93
    もしれない
    -0.86
    uerdo
    -0.84
    hedge
    -0.82
     duomen
    -0.81
     každé
    -0.81
    といいます
    -0.80
    ítése
    -0.80
    lacian
    -0.80
    設施
    -0.79
    POSITIVE LOGITS
     consiste
    1.02
    ündete
    1.00
     consisted
    0.96
     where
    0.91
    での
    0.91
     zuständig
    0.91
     trei
    0.90
     consists
    0.88
     consisting
    0.87
     contains
    0.87
    Activations Density 0.009%

    No Known Activations