© 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. 66437
    Prev
    Next
    INDEX
    Explanations

    Sheeran, verandas, and makeovers

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

    The neuron responds strongly to relatively rare or content-heavy tokens—especially proper names (like “Ed Sheeran” or “Reddy”) and other low-frequency, distinctive words.

    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
    at
    -3.34
    </u>
    -2.86
     $
    -2.72
    ↵↵
    -2.69
    de
    -2.67
    鸰
    -2.58
    𖥧
    -2.52
     почти
    -2.45
    
    -2.45
    }(\
    -2.39
    POSITIVE LOGITS
    ،
    2.89
    嫪
    2.77
    芣
    2.75
     anún
    2.58
    怍
    2.56
    鸶
    2.52
     Dinas
    2.44
     accustomed
    2.39
    再多
    2.38
    徜
    2.36
    Activations Density 0.004%

    No Known Activations