© 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. 22-GEMMASCOPE-RES-131K
    7. 67885
    Prev
    Next
    INDEX
    Explanations

    test code blocks

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

    The neuron fires on long runs of whitespace (indentation) and related punctuation (colons), i.e. it detects code‐indenting space tokens.

    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
    formatie
    -1.02
    还有什么
    -1.00
    onore
    -0.98
    摞
    -0.96
    這樣的
    -0.96
     många
    -0.96
    就是这样
    -0.93
     planten
    -0.93
    很多的
    -0.91
    biete
    -0.91
    POSITIVE LOGITS
     først
    1.23
     how
    1.09
     musst
    0.99
     жить
    0.96
     first
    0.93
    واع
    0.92
     variety
    0.91
     ギフト
    0.90
     do
    0.90
     случа
    0.89
    Activations Density 0.008%

    No Known Activations