© 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
    2. Gemma-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 7725
    Prev
    Next
    INDEX
    Explanations

    first strike / first person

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

    The neuron strongly activates on occurrences of the word “First” (or “first”) when it introduces ordinal or titled phrases (e.g. “First class,” “First Aid,” “First‐Year,” etc.).

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

    the word "First" when it appears at the beginning of a phrase or sentence, particularly in formal or commercial contexts.

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

    uses of the ordinal indicating primacy, especially when capitalized and appearing in hyphenated compounds or fixed expressions.

    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
     consecutive
    0.88
    th
    0.80
    最大的
    0.80
     consecutivos
    0.79
     najve
    0.77
    剩下
    0.77
     Consecutive
    0.74
     consecut
    0.74
     outra
    0.71
     grootste
    0.71
    POSITIVE LOGITS
     ever
    1.02
     responders
    0.95
    ever
    0.85
     responder
    0.85
     EVER
    0.85
     impressions
    0.83
     drafts
    0.75
    fruits
    0.74
     draft
    0.73
     timers
    0.73
    Activations Density 0.171%

    No Known Activations