© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 3481
    Prev
    Next
    INDEX
    Explanations

    sentence ending words

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

    The neuron specifically fires on the last meaningful word or content token at the end of a sentence or segment, effectively marking sentence/segment boundaries.

    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
     худо
    0.50
     валю
    0.48
    걍
    0.47
    ാലി
    0.46
     escrow
    0.46
    fting
    0.46
    یت
    0.46
     wyłącznie
    0.46
    ité
    0.46
    iteit
    0.45
    POSITIVE LOGITS
    ។
    0.57
    。...
    0.57
    。
    0.55
    Questo
    0.54
    。”
    0.54
    हमारे
    0.53
    。,
    0.52
     этому
    0.51
    喲
    0.51
     quinquina
    0.50
    Activations Density 0.095%

    No Known Activations