© 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. 6996
    Prev
    Next
    INDEX
    Explanations

    sentence terminators

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

    The neuron fires strongest over very long contiguous stretches of text—i.e. it acts as a length (or context‐window) detector rather than on any particular word or concept.

    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
     সমস্যাবলী
    0.94
    Fortunately
    0.94
    ؛
    0.90
     ؟
    0.86
    THERE
    0.85
    ^{\#}\
    0.84
    Caution
    0.84
    pgamma
    0.82
     જરૂ
    0.82
    Phenyl
    0.82
    POSITIVE LOGITS
     buổi
    0.74
     bunun
    0.72
     protest
    0.70
     şirk
    0.68
    他
    0.68
     ilk
    0.67
     기존
    0.66
     ül
    0.66
     pār
    0.66
     underworld
    0.66
    Activations Density 0.000%

    No Known Activations