© 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
    2. Gemma-3-27B-IT
    3. 10-GEMMASCOPE-2-TRANSCODER-262K
    4. 48257
    Prev
    Next
    INDEX
    Explanations

    The neuron is focused on ordinal numbers, particularly "3rd" and similar constructions, often linked to historical periods or dates like "century".Let's refine this.MAX_ACTIVATING_TOKENS: `3`, `th`, `3`, `th`, `3`, `th`, `th`, `th`, `th`TOKENS_AFTER_MAX_ACTIVATING_TOKEN: `7`, `century`, `0`, `century`, `6`, `century`, `century`, `to`, `century`, `Century`The pattern seems to be detecting numbers followed by 'th', and then often 'century' or digits followed by 'century'

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/transcoder_all/layer_10_width_262k_l0_small_affine
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Aryan
    0.55
    И
    0.51
    ע
    0.49
    О
    0.48
    Вер
    0.48
    お金
    0.47
    Во
    0.47
    А
    0.45
    Κ
    0.45
    Č
    0.44
    POSITIVE LOGITS
    ↵
    0.55
     függ
    0.47
    ).
    0.46
    ing
    0.46
    y
    0.45
    '
    0.45
     presup
    0.44
     inté
    0.44
     condu
    0.44
     thú
    0.43
    Activations Density 0.017%

    No Known Activations