© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 12910
    Prev
    Next
    INDEX
    Explanations

    destination directory, card payment

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

    The neuron fires on the very first “content” word of a new text segment—i.e. it zeroes in on paragraph‐ or section-initial tokens (often names, labels or technical identifiers that kick off each block).

    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
    conditions
    0.80
    ز
    0.79
    th
    0.76
    Tr
    0.76
    TH
    0.72
    Га
    0.71
    TO
    0.71
    Polit
    0.70
    قدم
    0.70
    c
    0.70
    POSITIVE LOGITS
     descubre
    1.02
     する
    1.01
     constate
    0.93
     apasion
    0.90
     ή
    0.86
     점에서
    0.85
     আপনি
    0.85
     próximo
    0.85
     fech
    0.85
     وهو
    0.84
    Activations Density 0.000%

    No Known Activations