© 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. 22-GEMMASCOPE-RES-131K
    7. 65729
    Prev
    Next
    INDEX
    Explanations

    code structure and punctuation

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

    The neuron strongly activates on comma tokens (the commas separating items or parameters 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_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
    Only
    -2.59
    现在的
    -2.47
    Maybe
    -2.34
    One
    -2.34
    自从
    -2.33
    Not
    -2.25
    Surprisingly
    -2.23
    　　-
    -2.20
     bolster
    -2.20
     stund
    -2.19
    POSITIVE LOGITS
    "'
    2.92
    
    2.72
     sabias
    2.66
    訁
    2.66
     involucrados
    2.64
    籥
    2.64
     protocolos
    2.59
    鷙
    2.56
    ,.
    2.56
     religiosos
    2.52
    Activations Density 0.014%

    No Known Activations