© 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. 13467
    Prev
    Next
    INDEX
    Explanations

    `que` followed by Spanish/Portuguese, `that` followed by English

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

    The neuron fires most strongly on low‐frequency “rare” words (including proper names, foreign words, and uncommon technical terms), i.e. tokens that occur infrequently in the training corpus.

    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
    ⛦
    -1.80
     clínico
    -1.80
     comprometido
    -1.79
    ⤞
    -1.78
     processos
    -1.75
    巨大的
    -1.74
     recopila
    -1.74
     relies
    -1.74
     дня
    -1.73
     reempla
    -1.73
    POSITIVE LOGITS
    ,
    2.47
    </h1>
    2.00
    Don
    1.95
    href
    1.93
    2
    1.88
    0
    1.85
    </h3>
    1.81
    Doesn
    1.68
     Some
    1.66
    9
    1.64
    Activations Density 0.003%

    No Known Activations