© 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. 24-GEMMASCOPE-2-RES-16K
    7. 5918
    Prev
    Next
    INDEX
    Explanations

    ` code ` function block

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

    This neuron fires on the lead‐in of news/press‐release style paragraphs—i.e. the first word (often a capitalized “The” or similar) at the start of a new section or dateline.

    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
    leyeceğiz
    0.55
    ették
    0.54
    ውነ
    0.53
    <unused470>
    0.53
    డాది
    0.52
    📽
    0.52
    tní
    0.51
    <unused1861>
    0.50
    <unused89>
    0.49
    iffent
    0.48
    POSITIVE LOGITS
     this
    0.61
     comes
    0.60
     Massachusetts
    0.56
     which
    0.55
     American
    0.55
     explores
    0.54
     describes
    0.54
     museums
    0.54
     bestselling
    0.54
     unassuming
    0.54
    Activations Density 0.271%

    No Known Activations