© 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-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 12180
    Prev
    Next
    INDEX
    Explanations

    Seraphina, Serenity, SRV

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

    The neuron fires on any token containing the character sequence “sar,” effectively detecting substrings like “Sar,” “sar,” or “SAR” within names or words.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    the token "Sar" when it appears as part of names (especially South Asian names like Sarita, Saracibar) or acronyms (like SARP).

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    tokens containing the substring “Sar,” especially at the start of proper names or acronyms.

    oai_token-act-pair · gpt-5Triggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Anomaly
    0.56
    ಿಂತ
    0.55
    র্কে
    0.54
    pecies
    0.54
    টু
    0.53
    rischen
    0.53
    酊
    0.53
    cept
    0.53
    raccoon
    0.52
     Handsome
    0.52
    POSITIVE LOGITS
    Ser
    0.68
    Sar
    0.62
    𒄀
    0.61
     Sar
    0.60
     सर्
    0.59
     சர
    0.58
     Ser
    0.55
    ября
    0.55
    热
    0.55
     sar
    0.55
    Activations Density 0.238%

    No Known Activations