© 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. 100181
    Prev
    Next
    INDEX
    Explanations

    spikey, scratchy, laggy

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

    The neuron strongly activates on subword tokens that form the ends of words (i.e. suffix fragments) rather than on full 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
    ↵
    -2.86
    o
    -2.53
    “
    -2.36
    ?
    -2.19
    </sup>
    -2.19
    由於
    -2.17
    {
    -2.14
     ''
    -2.13
    是
    -2.09
     lauded
    -2.05
    POSITIVE LOGITS
    ^{*},
    2.52
    釭
    2.30
     diese
    2.23
    ेशन
    2.14
    ！"
    2.08
    〜〜
    2.06
     obsessive
    2.02
    these
    2.00
    ዬ
    1.98
    好用
    1.97
    Activations Density 0.034%

    No Known Activations