© 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. 5305
    Prev
    Next
    INDEX
    Explanations

    'and' followed by conjunctions or list items

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

    The neuron fires on the connecting element or the first token of the second half in fixed two-part phrases (e.g. “mortar and pestle,” “Jack & Jill,” “cap and gown,” “keep and bear arms,” “hook-and-eye,” “slip-and-fall”).

    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
    ه
    0.50
    ة
    0.48
    ج
    0.46
    лог
    0.44
    تي
    0.44
    س
    0.44
    ஸ்
    0.41
    хара
    0.41
    ير
    0.41
    2
    0.40
    POSITIVE LOGITS
     and
    0.48
     precarious
    0.48
     deftly
    0.46
    一系列
    0.45
    ndash
    0.45
    hellip
    0.45
     drastically
    0.45
     fakat
    0.45
     versus
    0.44
     categor
    0.44
    Activations Density 0.013%

    No Known Activations