© 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. 1269
    Prev
    Next
    INDEX
    Explanations

    creating sections or new paragraphs

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

    The neuron is primarily firing on sentence‐final boundary tokens—i.e. it marks ends of sentences.

    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
     역할을
    0.85
     sourceL
    0.82
     veditabbo
    0.80
     và
    0.80
     compels
    0.79
     tích
    0.77
     afirmou
    0.77
    ✤
    0.77
     주제
    0.76
     способны
    0.76
    POSITIVE LOGITS
    IN
    0.80
    روش
    0.72
    Page
    0.70
    AR
    0.67
     اپ
    0.66
    こちらの
    0.66
    ET
    0.64
    ów
    0.64
    real
    0.64
    title
    0.63
    Activations Density 0.000%

    No Known Activations