© 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-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 8601
    Prev
    Next
    INDEX
    Explanations

    legal terms or phrasing

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

    The neuron is highly responsive to the very first meaningful token of each new paragraph or section header — in other words, it marks the opening word of a distinct text block.

    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_24_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
    دي
    0.92
    ع
    0.91
     куда
    0.90
    ྥ
    0.86
    да
    0.85
    ма
    0.84
     фирмы
    0.84
     logam
    0.83
     дочери
    0.81
    น์
    0.80
    POSITIVE LOGITS
     OPERATIONS
    0.82
     ECONOM
    0.81
     Rector
    0.79
     Headquarters
    0.77
    ⏩
    0.76
     PRODUCTION
    0.74
     Police
    0.73
     FAST
    0.72
     SUBSTITUTE
    0.71
     PARTS
    0.71
    Activations Density 0.000%

    No Known Activations