© 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. 392
    Prev
    Next
    INDEX
    Explanations

    sentence endings followed by pronouns

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

    This neuron fires most strongly on tokens appearing at the very end of a long input chunk, effectively marking the context-window boundary.

    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
     것입니다
    0.59
     möglicherweise
    0.48
    습니까
    0.47
    されます
    0.46
    とされる
    0.44
    これにより
    0.44
    =?";
    0.43
    함으로써
    0.43
     veroor
    0.42
    こうした
    0.41
    POSITIVE LOGITS
     teníamos
    0.79
     hicimos
    0.77
     ছিলাম
    0.77
     jsem
    0.76
     tuve
    0.75
     havde
    0.73
     jsme
    0.72
     माझ्या
    0.72
     আমার
    0.71
     मेरे
    0.70
    Activations Density 0.001%

    No Known Activations