© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 10482
    Prev
    Next
    INDEX
    Explanations

    first step in sequence

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

    The neuron strongly activates on the ordinal “first” when it introduces a numbered time or sequence element (e.g. “first day,” “first year,” “first step”).

    oai_token-act-pair · o4-miniTriggered 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
     처음
    0.82
     প্রথমবারের
    0.73
     weitere
    0.72
     سایر
    0.70
     різні
    0.68
     verschiedene
    0.65
     FURTHER
    0.65
    庸
    0.64
     सारी
    0.63
     первое
    0.63
    POSITIVE LOGITS
     onwards
    0.74
     installment
    0.71
     подряд
    0.71
     ever
    0.69
    जे
    0.69
    ของการ
    0.67
     instalment
    0.64
     arrivals
    0.63
    ever
    0.60
     onward
    0.55
    Activations Density 0.351%

    No Known Activations