© 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. 113789
    Prev
    Next
    INDEX
    Explanations

    closing bold tags

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

    triple single quotes (''') and markdown-style bold markers (**) used in documentation and code formatting.

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

    This neuron detects formatting or markup delimiters (e.g., Markdown `**`, wiki `'''`, HTML `<b>`/component tags) rather than natural language content.

    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
    ."
    -1.95
     that
    -1.84
    "));
    -1.80
     before
    -1.77
     \\
    -1.75
     when
    -1.74
     because
    -1.71
    :
    -1.71
     if
    -1.70
     during
    -1.66
    POSITIVE LOGITS
    </b>
    2.36
    </strong>
    2.23
    躋
    1.83
     zoude
    1.78
    はっきり
    1.76
    どうして
    1.74
    どのように
    1.73
    姈
    1.73
     corect
    1.66
    
    1.63
    Activations Density 0.059%

    No Known Activations