© 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. 10-GEMMASCOPE-RES-131K
    7. 103748
    Prev
    Next
    INDEX
    Explanations

    entrusted, strtolower, nofollow, Transactional, racycline, CrossOrigin

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

    words that end in "trust" or are related to trust, delegation, and assignment of responsibility.

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

    The neuron selectively activates on relatively long, multi-syllabic content words—especially formal or technical terms (e.g. “entrusted,” “Transactional,” “anthracycline,” “CrossOrigin”).

    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
    <td>
    -2.91
    </i>
    -2.89
    i
    -2.67
    9
    -2.47
    2
    -2.42
    The
    -2.42
    B
    -2.38
    that
    -2.36
    ：
    -2.28
    In
    -2.27
    POSITIVE LOGITS
    䊏
    2.86
    玙
    2.64
    䊐
    2.52
    最后由
    2.47
    卺
    2.44
    ﱠ
    2.38
    腚
    2.36
    ,’’
    2.30
    
    2.28
    豨
    2.27
    Activations Density 0.001%

    No Known Activations